Sei sulla pagina 1di 96

Aalborg Universitet

CFD Modelling and Experimental Testing of Thermal Calcination of Kaolinite Rich Clay
Particles - An Effort towards Green Concrete
Gebremariam, Abraham Teklay

Publication date:
2015

Document Version
Publisher's PDF, also known as Version of record

Link to publication from Aalborg University

Citation for published version (APA):


Gebremariam, A. T. (2015). CFD Modelling and Experimental Testing of Thermal Calcination of Kaolinite Rich
Clay Particles - An Effort towards Green Concrete. Department of Energy Technology, Aalborg University.

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

? Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
? You may not further distribute the material or use it for any profit-making activity or commercial gain
? You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us at vbn@aub.aau.dk providing details, and we will remove access to
the work immediately and investigate your claim.

Downloaded from vbn.aau.dk on: februar 08, 2018


CFD MODELING AND EXPERIMENTAL TESTING OF THERMAL
CALCINATION OF KAOLINITE RICH CLAY PARTICLES: AN EFFORT
TOWARDS GREEN CONCRETE

Abraham Teklay Gebremariam


Dissertation submitted to the faculty of Engineering and Science
at Aalborg University in partial fulfillment of the
requirements for the degree of

DOCTOR OF PHILOSOPHY

Aalborg University
Department of Energy Technology
Aalborg, Denmark
November 2015
ii

CFD modeling and experimental testing of thermal calcination of kaolinite rich clay
particles–An effort towards green concrete

Copyright Abraham
c Teklay, 2015

Printed in Denmark by Uniprint

ISBN: 978-87-92846-51-8

AALBORG UNIVERSITY
Department of Energy Technology
Pontoppidanstræde 101
Aalborg East, DK-9220
Denmark
“If we knew what we were doing, it wouldn’t be called research”

Albert Einstein.
Abstract

Cement industry is one of the major industrial emitters of greenhouse gases, generating
5-7% of the total anthropogenic CO2 emissions. Consequently, use of supplementary
cementitious materials (SCM) to replace part of the CO2 -intensive cement clinker is
an attractive way to mitigate CO2 emission from cement industry. SCMs based on
industrial byproducts like fly ashes and slags are subject to availability problems. Yet
clays are the most ubiquitous material on earth’s crust. Thus, properly calcined clays
are a very promising candidates as SCMs to produce green cements. Calcination at
inappropriately high temperatures or long retention time will not only waste energy but
also decrease the reactivity of the calcines due to possible recrystallization of the reactive
phase into a stable crystalline phase. Therefore, it is very crucial to achieve an in-depth
understanding of the calcination processes in a calciner and develop a useful tool that
can aid in the design of a smart clay calcination technology, which makes the major
objective of this study.

In this thesis, a numerical approach is mainly used to investigate the flash calcina-
tion of clay particles. A transient one-dimensional particle model which fully addresses
not only the particle-ambient flow interaction but also the intra-particle processes has
been successfully developed in a C++ program to examine calcination of clay parti-
cles suspended in a hot gas. The calcination process is also numerically studied using
gPROMS (a general PROcess Modeling System) software. The model results from both
C++ and gPROMS software show good similarity. Various experiments have been per-
formed to derive key kinetic data, to collect data from a gas suspension calciner (GSC),
and to characterize the calcines obtained under different calcination conditions, which
are either provided to the numerical model as inputs or as database for model validation.

The model is able to reliably predict the temperature and residence time at which a
given clay material attains optimum composition of the required material, metakaolinite.
For kaolinite rich clay particles with mean particle size of 13.74 µm in diameter, moderate
calcination temperatures (1173–1200 K) tend to display optimum amount of metakaol–
inite in a fraction of seconds with less risk of further phase transformation. High cal-
cination temperatures (>1300 K), however, deplete the amount of metakaolinite and
promote further recrystallization of metakaolinite into undesired mullite phase that in-
fluences pozzolanic property of calcines negatively. Different indicators have been used
to spot the optimum pozzolanic property of the calcined clay material, among which is
the density of calcines. By using the variation in density of calcines, an optimum resi-
dence time has been marked. At this time the calcines display a minimum density that
corresponds to the most dehydroxylated calcines. The behavior of flash calcined kaoli-
nite rich clays has also been examined experimentally. The composition and property
of calcines observed experimentally supports model prediction. The agreement between
model and experimental results confirms the validity of the model.

The optimum calcination parameters predicted in this study are crucial not only
to maximize the yield of the desired product but also minimize the energy consumption
during operation. Thus, the experimentally validated calcination model and simulation
results can aid in an improved understanding of clay calcination process and also new
conceptual design and optimization of clay calciners.
Dansk resume

Cementindustrien er en af de største udledere af drivhusgasser og producerer 5-7% af den


totale menneskeskabte CO2 emission. Som følge deraf er brug af alternative cementagtige
materialer (Supplementary Cementitious Materials - SCM) en attraktiv måde at erstatte
en del af de CO2 –intensive cement klinker og derved dæmpe udledningen af CO2 fra
cement industrien. SCMs baseret på industrielle biprodukter, såsom flyveaske og slagge,
er vanskelige, da tilgængeligheden af produkterne er meget begrænsede. Derimod er ler
det dominerende materiale i jordens skorpe, og derfor er korrekt kalcineret ler en meget
lovende kandidat til SCM for producktion af grøn cement. Lerkalcinering bør finde
sted indenfor et snævert temperatur- og tids-vindue: kalcinering ved uhensigtsmæssige
høje temperature eller ved for lang retentionstid vil ikke bare være spild af energi, men
også reducere produktets reaktionsevne, idet sandsynligheden for krystallisering af den
reaktive fase over i en stabil krystallisk fase herved stiger. Det er derfor ekstremt vigtigt
at opnå en dybdegående forståelse for kalcineringsprocessen for ler i en kalcinator, samt
at implementere denne i et brugbart værktøj, som kan medvirke i design processen af
“smart” ler kalcineringsteknologi, hvilket er hovedformålet med dette studie.

I afhandlingen er en numerisk tilgang hovedsageligt blevet brugt til at undersøge


flash kalcinering af ler partikler. En transient endimensional partikelmodel, som udførligt
kan håndtere såvel det partikel-nære flow som de intra-partikulære processer, er blevet
udviklet i et C++ program med det formål at undersøge kalcinering af lerpartikler
opslemmet i en varm gas. Kalcineringsprocessen er endvidere blevet undersøgt numerisk
ved brug af softwaren gPROMS (a general PROcess Modeling System) software. De
modellerede resultater fra både C++ og gPROMS viste fremragende overensstemmelse.
Et antal eksperimenter er blevet udført med henblik på at udlede de væsenligste kinetiske
data fra en suspensionskalciner (GSC) og til at karakerisere det kalcinerede ler opnået
under forskellige kalcineringsprocesser. De kinetiske data er derefter enten givet til den
numeriske model som input eller som en database for model validaering.
Modellen er i stand til pålideligt at kunne forudse temperatur og opholdstid, hvorved
et givet lermateriale opnår den optimale sammensætning af det ønskede materiale,
metakoalinit. For lerpartikler rige på kaolonit, med en gennemsnitlig størrelse på 13.74 µm
i diameter, opnåedes maksimal kaolinit-indhold nærmest øjeblikkeligt ved moderate
kalcineringstemperaturer på 1173-1200 K. Samtidig var dette en stabil tilstand, med
minimal risiko for videre fasetransformation. Høje kalcinerings temperature (>1300 K)
derimod gav meget lave kaolinit-niveauer og gav desuden anledning til øget krystalliser-
ing af metakaolinit hen imod den uønskede mullit fase, som påvirker det kalcinerede lers
pozzolaniske egenskaber i en negativ retning. Forskellige indikatorer er blevet benyttet
til at finde den optimale pozzolaniske egenskab af det kalcinerede ler materiale, blandt
hvilke er densiteten af den kalcinerede ler. Ved at benytte en variation i densiteten af
det kalcinerede ler, har det været muligt at fastslå en optimal opholdstid. Til dette
tidspunkt viser det kalcinerede ler en minimumsdensitet , hvilket svarer til det mest
hydroxylerede kalcinerede ler. Det flash kalcinerede og kaolinit-rige ler er også blevet
undersøgt eksperimentelt. Sammensætningen og egenskaber af det kalcinerede ler fra
eksperimenterne understøtter de forudsagte resultater fra modellen. Overensstemmelsen
imellem modellen og de eksperimentelle resultater bekræfter validiteten af modellen.

De optimale kalcineringsparametre fundet i dette studie er signifikante ikke bare


for at kunne maksimere afkastet af en det ønskede produkt, men også i forhold til at
minimere energiforbruget under drift. Således er det vist, at resultaterne fra den eksper-
imentelt validerede kalcineringsmodel og simuleringer kan medvirke til en forbedret
forståelse af ler kalcinering samt i et nyt conceptuel design og optimering af ler kalcina-
torer.
Acknowledgements

This work constitutes the effort, influence and inspiration of many individuals whom I
would like to acknowledge. First and above all, I praise and thank GOD who bestowed
us the power for learning, searching and performing.

Next, I would like to express my deepest gratitude to my principal supervisor


Associate Professor Chungen Yin for his invaluable time, encouragement and support
invested during the supervision of this project. I am also heartily thankful to my co-
supervisor Professor Lasse Rosendahl for his kind advice and discussion.

I am indebted to the persons who actively participated in this research. Therefore,


I would take this opportunity to thank Martin Bøjer, Anicka Adelsward, Lea Lindequist
and Klaus Hjuler for their helpful discussion and suggestions.

My appreciation and admiration also extends to FLSmidth A/S for providing office
and other facilities during my stay in Valby, Copenhagen. Also special thanks to the
people at FLSmidth research center in Dania who helped me during my laboratory
experiments. I am also thankful to Thomas Søren at the Department of Mechanical and
Production Technology for his support during TGA experiments. The same gratitude
goes to Marie, Lars and Samuel for reading and commenting on the thesis.

There are many more who deserve my gratitude, far too many to list here. In
general, I am truly grateful to all of my colleagues whom I have come to know during
my stay at the university and have spent many memorable fun all together.

My sincere thanks also go to project partners Aalborg Portland and Arhus univer-
sity (iNANO) for their discussion and feedback during the course of the project.

Lastly, I would like to express all my gratitude to my parents for their continuous
love and encouragement; in particular to my beloved wife Hilina and my son Tobias, for
their patience and understanding during the long hours spent at office.

viii
Contents

Abstract iv

Dansk resume vi

Acknowledgements viii

List of Figures xi

List of Tables xiii

Nomenclature xv

Mandatory page xix

1 Introduction 1
1.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objectives and scope of thesis . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Contribution of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Thesis outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Background and literature review 7


2.1 Production of cement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Impact of cement production on the environment . . . . . . . . . . 9
2.2 Supplementary Cementitious Materials (SCMs) . . . . . . . . . . . . . . . 10
2.2.1 The need for SCMs . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Pozzolanic reactivity . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Kaolinite . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.1 Calcination of kaolinite . . . . . . . . . . . . . . . . . . . . . . . . 14
2.3.2 Kinetics of dehydroxylation . . . . . . . . . . . . . . . . . . . . . . 16
2.4 Modeling thermal calcination of kaolinite . . . . . . . . . . . . . . . . . . 17
2.5 Experimental study on flash calcination of kaolinite . . . . . . . . . . . . . 20

3 Single particle calcination model 23


3.1 Fundamentals of CFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.2 Description of the clay particle calcination model . . . . . . . . . . . . . . 24

ix
Contents x

3.2.1 Main assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . 25


3.2.2 Governing equations . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2.3 Boundary conditions . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2.4 Initial conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3 Discretization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.1 Spatial discretization . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.2 Temporal discretization . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3.3 Final discretized equations . . . . . . . . . . . . . . . . . . . . . . 31
3.4 Iterative procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.5 Modeling tools: C++ and gPROMS ModelBuilder . . . . . . . . . . . . . 35

4 Experimental study 37
4.1 Study of kinetic parameters . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.1.1 Thermogravimetric study . . . . . . . . . . . . . . . . . . . . . . . 39
4.2 Gas Suspension Calciner (GSC) . . . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Characterization test of the raw feed and the calcines . . . . . . . . . . . 42
4.3.1 Mineralogical composition and PSD . . . . . . . . . . . . . . . . . 43
4.3.2 Density and surface area . . . . . . . . . . . . . . . . . . . . . . . . 43
4.3.3 Degree of dehydroxylation . . . . . . . . . . . . . . . . . . . . . . . 44
4.3.4 Composition of phases . . . . . . . . . . . . . . . . . . . . . . . . . 46

5 Result and discussion 49


5.1 Kinetic parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2 Experimental and modeling results of flash calcination in the GSC: A
comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5.3 Detailed model prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3.1 Baseline case results . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.3.2 Sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

6 Conclusion 63
6.1 Final remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

Bibliography 67
List of Figures

2.1 The production process of Portland cement showing the major pollution
sites of carbon dioxide and particulate matter(PM). . . . . . . . . . . . . 8
2.2 The global greenhouse gases emission in the year 2000 and contribution
of cement industry [Rehan and Nehdi, 2005]. . . . . . . . . . . . . . . . . 9
2.3 The arrangement of tetrahedral silica and octahedral alumina layers to
form kaolinite structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.1 A sketch of 1D clay particle calcination model . . . . . . . . . . . . . . . . 25


3.2 Discretization of the one dimensional clay particle . . . . . . . . . . . . . 31
3.3 Overall algorithm for the sequence of operations in the problem . . . . . . 34
3.4 A screenshot of gPROMS interface . . . . . . . . . . . . . . . . . . . . . . 36

4.1 The shift in temperature during dehydroxylation of kaolinite rich clay at


different heating rates, θ (◦ C/min) . . . . . . . . . . . . . . . . . . . . . . 38
4.2 A representative thermogram for an experiment done at a heating rate of
20 K/min . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 TGA equipment used during study . . . . . . . . . . . . . . . . . . . . . . 40
4.4 Plot of ln(θ/Tp2 ) versus 1/Tp to determine kinetic parameters . . . . . . . 41
4.5 A schematic sketch of the gas suspension calciner along with the position
of secondary burners (SB) and thermocouples (TC). . . . . . . . . . . . . 42
4.6 Particle size distribution of the feed and calcined clay materials. . . . . . 44
4.7 LECO RC612 Multiphase Carbon and Hydrogen/moisture analyzer. . . . 45
4.8 The amount of vapor in terms of weight fraction, as it is shown in LECO
RC612: (a) Feed (b) calcined clay at 1273 K (MK1273WB ) . . . . . . . . 45
4.9 A graphical plot of the intensity of vapor during LECO RC612 thermal
analysis for the feed and calcined clay products. . . . . . . . . . . . . . . . 46
4.10 The PANalytical CubiX PRO X-ray diffractometer used in this study. . . 47
4.11 XRD pattern of the as received clay sample and its calcines at different
calcination temperature. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

5.1 The influence of kinetic parameters on the rate constant . . . . . . . . . . 50


5.2 Comparison between GSC tests and model predictions: density of the
clay samples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.3 Comparison between GSC tests and model predictions: Degree of dehy-
droxylation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.4 The comparison between C++ and gPROMS results (A) temperature
profile at particle center/surface (B) model predicted conversion profile . . 54
5.5 Model predicted profiles in temperature and water vapor at particle cen-
ter/surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55

xi
List of Figures xii

5.6 Model predicted density of solid species and conversion of the particle . . 56
5.7 The changes in density of kaolinite as a function of degree of dehydrox-
ylation during calcination: comparison between experiment, empirical
correlation and model prediction. . . . . . . . . . . . . . . . . . . . . . . . 57
5.8 The variation in model predicted density of the calcined product as a
function of time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
5.9 Regime diagram for convection, diffusion and chemical kinetics repro-
duced from Literature [Lopes et al., 2009] showing a specific region that
belong to the calcination process of kaolinite clay (dashed line) . . . . . . 59
5.10 Sensitivity of calcination temperature and time on (A) model predicted
conversion (B) model predicted mass fraction of metakaolinite . . . . . . . 60
5.11 Sensitivity of particle size on model predicted conversion . . . . . . . . . . 61
5.12 The sensitivity of kinetic parameters based on literature and their impact
on the conversion profile of the model . . . . . . . . . . . . . . . . . . . . 62
List of Tables

2.1 Effects of SCMs on the properties of hardened concrete . . . . . . . . . . 12


2.2 A summary of kinetic parameters of kaolinite dehydroxylation from liter-
ature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.1 The expression of reaction rate and kinetic parameters used in the model 26
3.2 The Physical and transport properties and their expression in the model . 27
3.3 The expressions for solid species source/sink terms during calcination of
kaolinite clays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.4 Temperature dependent heat capacity of solid species . . . . . . . . . . . . 28
3.5 A solution to the discretized energy equation . . . . . . . . . . . . . . . . 32
3.6 A solution to the discretized transport equation of gases . . . . . . . . . . 33
3.7 A solution to the discretized pressure-correction equation. . . . . . . . . . 33

4.1 Experimental peak characteristic values from Experiment 1. . . . . . . . . 40


4.2 Experimental peak characteristic values from Experiment 2. . . . . . . . . 40
4.3 Calcination conditions of the GSC and details of sample ID . . . . . . . . 42
4.4 Chemical composition of kaolinite rich clay sample . . . . . . . . . . . . . 43

5.1 A summary of kinetic parameters for kaolinite dehydroxylation from lit-


erature and experimental values of this study . . . . . . . . . . . . . . . . 50
5.2 Experimentally estimated and model-predicted composition of clay samples 53
5.3 The initial properties of kaolinite clay and calcination conditions . . . . . 54
5.4 A summary of kinetic parameters for kaolinite dehydroxylation from lit-
erature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

xiii
Nomenclature

Ak Frequency factor of k th reaction s−1


Asurf Particle surface area m2
Bi Biot number -
Cp Specific heat kJ/(kg·K)
D Diffusion coefficient in pores m2 /s
DAB Molecular diffusivity of gase A in B m2 /s
Deff Effective Diffusivity m2 /s
Dj,m Diffusivity of species j in gas mixture m2 /s
dp Pore diameter m
Ek Activation energy of k th reaction kJ/mol
4Hk Heat of k th reaction kJ/kg
h Sensible enthalpy of gas mixture kJ/kg
hi,j Sensible enthalpy of solid/gas species kJ/kg
hm Mass transfer coefficient m/s
hT Heat transfer coefficient W/(m2 ·K)
keff Effective conductivity W/(m·K)
kg Gas conductivity W/(m·K)
ks Conductivity of solid W/(m·K)
M Wi Molecular weight of the ith species kg/kmol
n Order of reaction -
Nu Nusselt number -
P Pressure Pa
Patm Atmospheric pressure Pa
P em Peclet mass transfer number -
Pr Prandtl number -
ṙk Rate of k th reaction kg/m3 ·s
R Universal gas constant kJ/(mol·K)
Re Reynolds number -
Rp Particle radius m
Sc Schmidt number -
xv
Nomenclature xvi

Sh Sherwood number -
Sh Source/sink term in the energy equation W/m3
Sg Source/sink term in the continuity equation kg/(m3 ·s)
Sy,j Source/sink term in species transport equation kg/(m3 ·s)
t Time s
T Temperature K
Tf ilm Reference temperature of gas film K
Trad Radiation temperature K
Ts Particle surface temperature K
T∞ Ambient gas temperature K
U Physical velocity of gas m/s
V Particle volume m3
W1/2 Half width of a peak -
Yj Mass fraction of species j -
Yj,f ilm reference mass fraction of species j in the gas film -
Yj,s Mass fraction of species j at particle surface -
Yj,∞ Mass fraction of species j in the ambient gas -

Greek letters

α Degree of dehydroxylation -
β Mass ratio of crystal water in kaolinite -
ε Porosity -
η Permeability m2
θ Heating rate K/s
µ Dynamic viscosity kg/(m·s)
ρi Density of ith species kg/m3
ρg Gas density kg/m3
σ Stefan-Boltzmann constant W/(m2 K4 )
τ Tortuosity -
φ Thiele number -
ω Particle emissivity -

Subscripts

cr Cristobalite
fw Free water
kl Kaolinite
mk Metakaolinite
Nomenclature xvii

mu Mullite
sil Silica
sp Spinel

Abbreviations

ACI American Concrete Institute


ASR Alkali-Silica Reactions
BDNLSOL Block Decomposition Non Linear SOLver
CFD Computational Fluid Dynamics
DAE Differencial-Algebraic Equation
DASOLV Differencial-Algebraic equation SOLver
DSC Differential Scanning Calorimetry
DTA Differential Thermal Analysis
DTG Differential Thermo Gravimetry
FVM Finite Volume Method
GHG Green House Gas
gPROMS general PROcess Modeling System
LECO Laboratory Equipments COmpany
ODE Ordinary Differential Equation
PDE Partial Differential Equation
SCM Supplementary Cementitious Material
TDMA Tri Diagonal Matrix Algorithm
TGA Thermo Gravimetric Analysis
Mandatory page
Thesis title : CFD modeling and experimental testing of thermal calcination of kaoli-
nite rich clay particles : An effort towards green concrete
Name of PhD student : Abraham Teklay Gebremariam
Name and title of supervisors : Chungen Yin, Associate Professor; Lasse Rosendahl,
Professor
List of published papers:
Paper 1: Abraham Teklay; Chungen Yin; Lasse Rosendahl; Martin Bøjer.Calcination of
kaolinite clay particles for cement production: A modeling study. Cement and Concrete
Research 61-62 (2014), 11-19

Paper 2: Abraham Teklay; Chungen Yin; Lasse Rosendahl; Lea Lindequist Køhler. Exper-
imental and modeling study of flash calcination of kaolinite rich clay particles in a gas
suspension calciner. Applied Clay Science 103 (2015), 10-19.

Paper 3: Abraham Teklay; Chungen Yin; Lasse Rosendahl. Flash calcination of kaolinite
rich clay and impact of process conditions on the quality of the calcines: A way to re-
duce CO2 footprint from cement industry. (doi:10.1016/j.apenergy.2015.04.127, Applied
Energy, 2015)

Conference Contributions:
• Abraham Teklay; Chungen Yin; Lasse Rosendahl. Modeling of calcination of single
kaolinitic clay particle. Presented at the 8th International Conference on Multifluid
Flow (ICMF2013), Jeju, Korea, 2013
• Abraham Teklay; Martin Bøjer; Anicka Adelsward; Chungen Yin; Lasse Rosendahl. Sim-
ulation of flash dehydroxylation of clay particle using gPROMS: A move towards green
concrete. Proceedings in The 6th International Conference on Applied Energy, Taiwan,
2014; Energy procedia 61 (2014), 556-559

This present report combined with the above listed scientific papers has been submitted
for assessment in partial fulfilment of the PhD degree. The scientific papers are not
included in this version due to copyright issues. Detailed publication information is
provided above and the interested reader is referred to the original published papers. As
part of the assessment, co-author statements have been made available to the assessment
committee and are also available at the Faculty of Engineering and Science, Aalborg
University.

xix
Chapter 1

Introduction

This chapter gives an overview of the significance of the study and defines the objectives
and scope of the thesis. It also presents the methodology and approach used to arrive at
the final conclusions, followed by contribution of the study.

1.1 Overview

Fueled by the current environmental concern and the ever increasing demand for cement,
the present emission trends in cement industry need to be halted. The use of supple-
mentary cementitious materials to offset a portion of cement powder in concrete has got
paramount significance in lowering the environmental cost associated with the emission
of greenhouse gases from cement production. With this respect, the current study finds
a method of optimizing the pozzolanic reactivity of kaolinite rich clay particles that can
be used as SCMs. Kaolinite being one of the most abundant clay mineral, its excellent
pozzolanic properties has drawn renewed attention in cement and concrete industry.
Calcination of kaolinite is among the popular methods to produce the metastable phase
called metakaolinite, which has a particular mineralogical properties that can be utilized
as a mineral admixture in cement and concrete production.

Well-calcined clay products especially metakaolines have shown pozzolanic activity equal
to or higher than the well-known artificial pozzolans such as; fly ash, silica fume and
blast furnace slag. The fact that most clay minerals are compositionally suitable to be
used as pozzolanic materials is the starting point of the investigation.

1
Chapter 1. Introduction 2

1.2 Objectives and scope of thesis

One of the key visions of this project is to elaborate an innovative processes and know-
how needed to convert the locally available clays into high-quality SCM products. The
pozzolanic reactivity of such clays largely depends on the pyroprocessing or calcination
conditions (e.g., temperature, residence time of clay particles in the calciner), as well
as clay properties (e.g., type of clay, particle size distribution, morphology). With
this respect, the key objective of this study is to find out how to obtain clay particles
with the desired pozzolanic properties by smart calcination or pyroprocessing; so that
these reactive clay products could be used as a partial clinker substitute to mitigate
the emission of CO2 from cement industries. While Modeling is the main workhorse of
the study, supplemental laboratory tests and experiments are key inputs to validate the
model.

The main scope of this PhD study is therefore:

• Develop a calcination model for single clay particle, which is able to simulate the
heat transfer process and simultaneous changes in composition during thermal
calcination

• Predict the spatial and temporal evolution of temperature and composition inside
a clay particle of different size calcined under different conditions

• Validate the model experimentally by performing experiments in a pilot scale gas


suspension calciner

• Propose optimum calcination conditions (temperature and residence time) that


can be used to turn the clay material into useful SCM product

1.3 Methodology

This thesis focuses on the development of a numerical model to study the calcination
process of kaolinite rich clays, and experimental examination of the clay material using
TGA and pilot scale gas suspension calciner. The general approach is described briefly
as follows:

Numerical model

One-dimensional transient particle model has been developed using C++ and gPROMS
softwares. The major intention of using gPROMS software is to develop the calcination
Chapter 1. Introduction 3

model which later could be used to simulate results in a tubular reactor or combine
the gPROMS calcination model with CFD to run simulation of gas suspension calciner.
However, the use of gPROMS is suspended due to adjustments made by the project
consortium. The results generated by gPROMS are used as a verification effort for the
C++ results.

Generally, the model addresses the conversion of a kaolinite particle when inserted into
a hot gas atmosphere. The process can be understood by considering a porous kaolinite
clay particle, when suspended in the hot gas, heat is transferred to the particle surface
by convection and radiation. As heat penetrates into the surface, different reactions and
phase transformations commence causing the release of water vapor through the porous
surface. In order to address this phenomena numerically, the particle has been discretized
into a number of spherical cells, on each of which mass, momentum, energy and species
conservation equations are numerically solved by using the finite volume method. The
model reliably predicts the conversion and simultaneous changes in composition.

Experimental study

A number of experiments are carried out either to provide an input data to the model
or experimental evidences for model validation. These experiments cover a wide scope
of tests ranging from TGA to comprehensive flash calcination tests in a gas suspension
calciner, including various characterization methods of the feed and the calcines.

TGA experiments are performed to collect kinetic parameters (E and A) which are vital
inputs to the model. TGA tests are also used as part of model validation, where the
conversion of the clay particle is compared with model prediction.

Calcination experiments are performed in a pilot scale gas suspension calciner, where
flash calcines have been produced at different calcination conditions inside the GSC.
The composition and property of the calcines has been further analyzed.

Characterization of the feed and calcines have been carried out using several methods.
The composition and degree of dehydroxylation for the feed and calcines are investigated
using X-ray diffraction (XRD) and LECO thermal analyzer, respectively. The amount of
phases such as kaolinite, metakaolinite and mullite are compared with model predicted
ones. Other relevant properties such as the PSD, the density and specific surface area
of the feed and calcines are examined using laser diffraction, Gas Pycnometer and BET
method, respectively. Experimental outputs such as the degree of dehydroxylation and
the specific density are used to validate the model results.
Chapter 1. Introduction 4

1.4 Contribution of thesis

The contribution of this thesis is largely to elaborate an innovative process and know-
how needed to convert the locally available materials, such as clays, into high-quality
SCM products. The reactivity of clays largely depends on calcination conditions (e.g.,
temperature, residence time of clay particles in the calciner), as well as clay properties
(e.g., type of clay, particle size distribution, morphology). The fundamental knowledge
on these effects is understood by modeling the clay particle and simulating at different
calcination conditions. With this regard, the major contributions of this thesis are:

• A comprehensive calcination model for kaolinite particle has been developed. The
model considers several reactions such as dehydration, dehydroxylation and various
phase transformations that may occur when the clay particle is heated.

• Optimum calcination path for the conversion of kaolinite to metakaolinite is pre-


dicted. Consequently, the time-temperature-transformation curve is derived.

• The composition of calcines has been experimentally characterized after the clay
material has been being exposed to different calcination conditions in the GSC.
The amount of different phases has been quantified with the aid of experimental
techniques. Experimental results show reasonable agreement with model predic-
tion.

• A comprehensive model-based sensitivity analysis of the impacts of the most im-


portant calcination conditions (such as, temperature and time); and clay particle
properties (such as, particle size and kinetic parameters) has been carried out.
Their impact on the property of the final product has been explained.

1.5 Thesis outline

This thesis is prepared as a collection of scientific papers produced during the entire
PhD period. It is composed of 6 chapters which are described as follows;

Chapter 1 gives a general overview on the study and defines the scope and objectives
of the thesis. It also lists the contribution of the thesis.

Chapter 2 presents a solid background on the production of cement and the significance
of SCMs in alleviating CO2 from cement industries. It also presents literature review
on the calcination of clays, specifically on kaolinite rich clays. Finally, it reviews the
available literature on modeling and experimental study on calcination of kaolinite rich
clay particles.
Chapter 1. Introduction 5

Following the background and literature review, the details of a single particle calcination
model are extensively discussed in Chapter 3. The mathematical approach and major
assumptions are described at the beginning, then the governing equations and methods
of discretization are explained.

Chapter 4 describes the experimental section where several tests are performed in order
to determine the kinetic parameters of clay calcination, which are the main inputs of
the model. Calcination tests in a pilot scale gas suspension calciner are presented along
with characterization method of the calcines.

Chapter 5 is a summary of result and discussions. In this chapter, the major results
of this project are discussed in relation to the objectives of the study and the existing
literature. A brief discussion on model validation is made. Model predicted results are
also discussed and the impacts of key factors in clay calcination (e.g., particle size, gas
temperature and reaction kinetics) are also examined through a model-based sensitivity
analysis.

Finally, the main conclusion and recommendations for future work are provided in
Chapter 6.
Chapter 2

Background and literature review

This chapter presents the background on the production of cement and its impact on
the environment. It also highlights the advantage of using of SCMs in reducing CO2
footprint from cement industry. Finally, the efforts reported in literature to understand
the calcination of kaolinite rich clays either by using models or through experimental
study are reviewed.

2.1 Production of cement

Production of cement generally involves the following broad stages: quarrying and pro-
cessing of raw materials; pyroprocessing to produce clinker; blending and grinding of
clinker to cement and finally storage and packing.

Generally the raw materials are a mixture of minerals containing calcareous, siliceous,
argillaceous, and ferriferous materials that mainly include limestone, marl and shale
or clay, which are extracted from quarries. These materials are primarily crushed and
milled before they are transported to the cement plant for storage and further condi-
tioning. The raw materials, in controlled proportions, are ground and mixed together to
form a homogeneous blend in fineness and composition. Figure 2.1 provides a typical
production flowchart of Portland cement.

The minerals are transferred into the kiln system where the raw mix is fused at high
temperatures to alter into new minerals with hydraulic properties. Prior to introducing
the raw mix into the rotary kiln, it is pre-heated and calcined at the pre-calciners where
the majority of heat source comes from the waste heat of rotary kiln.

7
Chapter 2. Background and literature review 8

Figure 2.1: The production process of Portland cement showing the major pollution
sites of carbon dioxide and particulate matter(PM).

The heart of the Portland cement manufacturing process is the pyroprocessing stage
where the chemistry of the cement manufacturing process begins. This process trans-
forms the raw mix into clinker through the process called calcination. At this stage of
the process, calcium carbonates is decomposed to yield calcium oxide (CaO) and CO2 .
This is followed by the clinkering process at the rotary kiln where the raw material mix
enters the kiln at the elevated end in a counter current manner to the flow of fuels. As
the feed materials are continuously and slowly moved down the kiln, the raw materi-
als are changed to cementitious or hydraulic mineral called clinker, which is a product
of high temperature reactions (typically 1673 − 1773K) between calcium oxide, silica,
alumina, and ferrous oxides. The hot clinker then falls onto a grate cooler where it is
cooled rapidly by air.

The final step in the manufacturing of Portland cement is blending and grinding opera-
tions that transform clinker to finished Portland cement. Up to 5% gypsum or natural
anhydrite is added to the clinker during grinding to control the cement setting time. At
this stage, addition of other minerals (blast furnace slag, metakaolinite, fly ash or silica
fume) could be accomplished to reduce the CO2 footprint from the industry.
Chapter 2. Background and literature review 9

2.1.1 Impact of cement production on the environment

Most emissions that come out of cement industry are from the kiln system. The main
constituents of the exit gas are oxides of nitrogen, oxides of sulfur and oxides of carbon
generated due to process related chemical reactions and combustion of fuels. The emis-
sion of these gases is environmentally damaging. CO2 being the major constituent of
the exit gas, it is produced by two mechanisms:
1. Calcination of limestone or other calcareous materials at high temperature.

CaCO3 → CaO + CO2


1kg 0.56kg 0.44kg
2. Combustion of fuel during pyroprocessing that releases substantial quantities of CO2 .

As a result, cement industry is associated with the emission of large amounts of CO2
from both calcination and combustion processes. Every kilogram of Portland cement
generates 0.73–0.99 kilogram of CO2 which comprises approximately 5-7% of total an-
thropogenic CO2 emissions [Mehta, 2002]. When all greenhouse gas emissions gener-
ated by human activities are considered, the cement industry is responsible for approxi-
mately 3% of global emissions [Humphreys and Mahasenan, 2002; Mehta, 2002]. Figure
2.2 describes the global greenhouse gas emissions and contribution of cement industry
[Humphreys and Mahasenan, 2002; Rehan and Nehdi, 2005].

Global Greenhouse Gas (GHG) emission interms of CO2 equivalents [1Gt≈109 tonnes]

Other GHGs
14.8 Gt (34%)

Deforestation Calcination
3.94 Gt (9%) ~50%

Cement,
1.4Gt (3%)
Fuel
~40%

Fossil fuel
23.9 Gt (54%)
Transport and
electricity ~10%

Figure 2.2: The global greenhouse gases emission in the year 2000 and contribution
of cement industry [Rehan and Nehdi, 2005].
Chapter 2. Background and literature review 10

Like many other industrial activities, the cement industry needs to strictly comply with
national and international emission abatement regulations. Moreover, with the drastic
increase in Portland cement production foreseen in the future, the current cement indus-
try is facing challenges of producing more sustainable and less energy intensive products
without sacrificing the mechanical performance or durability of the end product. As a
result, different strategies have been forwarded to alleviate CO2 emission from cement
industry [IEA, 2009], among them are, the use of alternative fuels, use of carbon capture
and storage strategies and use of SCMs.

In this study the use of SCMs is of main interest, and hence, the use of natural or
artificial pozzolans such as calcined clays are considered as a useful admixtures of Port-
land cement. The potential future of clay materials as SCM is not only because of its
availability, but also due to its environmental advantage.

The use of metakaolinite as SCMs has been extensively studied in the literature [Arikan
et al., 2009; Cassagnabère et al., 2010; He et al., 1994; Khatib et al., 2012; Said-Mansour
et al., 2011; Salvador, 1995; Shvarzman et al., 2003; Tironi et al., 2012] and found that
partial substitution of metakaolinite (up to ∼ 20% of clinker) plays a significant role
in the production of green concretes without compromising the mechanical property
and durability of the blended concrete. However, proper thermal activation of kaolinite
clay is critical, because the above mentioned benefits are directly related to the amount
metakaolinite in the calcined product.

2.2 Supplementary Cementitious Materials (SCMs)

The use of pozzolana along with Portland cement was originally practiced to reduce the
cost of Portland cement. However, the mixture of pozzolana and Portland cement has
got other interesting benefits such as inhibiting or suppressing the alkali-silica reactions
that improve the durability of cement. In this regard, the construction of Friant dam in
California in 1940 was a typical example proving the benefits of blending cement with
natural pozzolan (Pumicite) to reduce the deterioration of concrete due to expansion
[Hewlett, 1998].

The use of metakaolin blended concrete has also been practiced since 1960’s to mitigate
alkali-silica reactions and improve durability of concrete; with this regard, the construc-
tion of Jupia Dam in Brazil in 1962 using metakaolinite as a partial substitute of cement
was a great success [Li et al., 2010].

The knowledge and the practice of substituting part of cement with SCMs has long been
exercised for several decades, and still it is under investigation. One can easily perceive
Chapter 2. Background and literature review 11

the complexity of the subject as it still attracts the attention of many researchers and
business firms.

SCMs are a group of materials that show either hydraulic or pozzolanic behavior, in
such a way that it can set and harden in the presence of water by forming cementitious
products in a hydration reaction [Massazza, 1993]. This group of materials embraces a
large number of materials which vary widely in terms of origin, mineralogical composi-
tion, and typical particle characteristics. Typical examples of such materials are fly ash,
silica fume, blast furnace slag, metakaolin, rice husk ash and natural pozzolans.

SCMs could be categorized in two broad distinctions: materials of natural origin (nat-
ural SCMs) and materials of man-made or artificial origin (artificial SCMs). Natural
SCMs consist materials that can be used in their naturally occurring form after being
conditioned by sieving and grinding processes. Typical examples are volcanic ash or
pumicite, shales, tuffs, and some diatomaceous earth.

Artificial SCMs includes materials which have undergone structural modifications as


a consequence of manufacturing or production processes. They can be produced de-
liberately by thermal method, for instance by thermal activation of kaolinite to obtain
metakaolin, or can be obtained as waste or by-products from high temperature processes
such as blast furnace slags, fly ashes or silica fume.

SCMs such as fly ash, slag and calcined clays are the most promising materials as cement
admixture. Among all SCMs, those based on calcined clay are the future materials as
cement admixture; primarily because of their availability and additionally they modify
the performance and durability of a concrete [Cassagnabère et al., 2010; Sabir et al.,
2001].

Today, artificial SCMs such as metakaolinite are enjoying a renaissance as SCMs in ce-
ment industry and may replace part of the clinker not only to mitigate CO2 footprints
from cement production but also to enhance the performance of the concrete by decreas-
ing the deleterious alkali-silica reactions [Cassagnabère et al., 2010; Lothenbach et al.,
2011; Sabir et al., 2001].

2.2.1 The need for SCMs

There has been a continuous search for material admixtures that can improve the work-
ability, strength and durability of a concrete. Recently, due to the strict air pollution
controls and regulations imposed on cement industry, an alternative material search has
primarily focused on materials that can be utilized to reduce CO2 footprint from cement
industry.
Chapter 2. Background and literature review 12

In addition to the environmental benefits, some of the broad benefits of incorporating


SCMs as partial cement replacement are; increased compressive strength, reduced segre-
gation, reduced permeability, mitigation of alkali-silica reactions and enhanced workabil-
ity. It has also been shown that using multiple SCMs as a cement admixture appear to
have synergistic effect in the concrete [Shehata and Thomas, 2002; Thomas et al., 1999].
This is because of the chemical and physical differences of each SCM additive stimu-
late different reactions in the concrete, granting a unique property of high-performance
concretes. Table 2.1 presents the general effect of SCMs on the property of hardened
concrete [Dam, 2013].

Table 2.1: Effects of SCMs on the properties of hardened concrete

Property Fly ash(Type C) Slag Silica fume Metakaoline


Early strength ↔ ↓ ↑↑ ↑↑
Long term strength ↑ ↑ ↑↑ ↑↑
Permeability ↓ ↓ ↓↓ ↓↓
Chloride ingress ↓ ↓ ↓↓ ↓↓
ASR l ↓↓ ↓ ↓
Sulfate resistance l ↑↑ ↑ ↑
Freezing and thawing ↔ ↔ ↔ ↔
Abrasion resistance ↔ ↔ ↔ ↔
Drying shrinkage ↔ ↔ ↔ ↔
• KEY
↓ Reduced
↓↓ Significantly reduced
↑ Increased
↑↑ Significantly increased
↔ No significant change
l Effect varies

One major concern regarding the use of SCMs may be the initial cost associated with
production the SCM-cement blend. Specially using metakaolinite as SCM may surge the
overall cost of the concrete, as calcination of this material incurs additional cost on the
blended cement. Usually the initial cost of a concrete containing metakaolinite additive
is expected to increase. The real economic savings are obtained over the lifecycle, as the
enhancements in ultimate strength, and durability often result an improved long-term
performance.

2.2.2 Pozzolanic reactivity

A Pozzolan is generally defined in ACI 116R as “..a siliceous or siliceous and aluminous
material which, in itself, possesses little or no cementitious value but which will, in finely
divided form and in the presence of moisture, react chemically with calcium hydroxide
(lime) at ordinary temperature to form compounds possessing cementitious properties”
Chapter 2. Background and literature review 13

[Tikalsky et al., 2001]. Hence, the term pozzolana includes all those inorganic materials,
either natural or artificial, which harden in water when mixed with calcium hydroxide.
Generally, it embraces a large number of very different materials in terms of origin,
composition and structure.

Pozzolanic reactivity, therefore, covers all reactions occurring among the active con-
stituents of pozzolanas, lime and water. Pozzolanic reactivity of SCMs is characterized
by the reaction between siliceous or aluminosiliceous material in the SCM with calcium
hydroxide (a reaction product from the hydration of Portland cement), forming calcium
silicate hydrate and other cementitious compounds that have a positive impact on the
long-term properties of the hardened concrete.

2.3 Kaolinite

Kaolin is a fine, white, clay mineral that has traditionally been used in the manufactur-
ing of porcelain. Kaolinite is the mineralogical term that is applicable to kaolin clays
composed of hydrated aluminum silicates. Its structural formula is Al2 Si2 O5 (OH)4 with
theoretical composition of 46.54% SiO2 , 39.5% Al2 O3 and 13.96% H2 O. The shape of a
perfectly ordered kaolinite crystal is pseudo-hexagonal, but its crystallinity may range
from a highly crystalline to a poorly ordered crystal [Prasad et al., 1991].

Kaolinite is the most prominent 1:1 type of phyllosilicate clay mineral. It has flake(flat)
type morphology due to the arrangement of atoms in the structure. The basic struc-
ture is made through the arrangement of two layers: the tetrahedral and octahedral
layer. Each tetrahedron consists Si4+ cation coordinated to four oxygen atoms that are
linked to adjacent tetrahedra by sharing three oxygen atoms at the corner; while the
octahedron consist Al3+ cations surrounded by six hydroxyl groups. The free corners
of the tetrahedral sheet connect with the octahedral sheets to form a common plane
or a single layered structure. The sheet of atoms are stacked on top of each other in
Tetrahedral-Octahedral (TO) fashion to form a layered structure. These layers, in turn,
are held together by hydrogen bonding between hydroxyls of the octahedral layer and
oxygen of the tetrahedral layer to form a repeating pattern of kaolinite clay structure as
shown in Figure 2.3.

Three fourth of the hydroxyl groups in the kaolinite structure lie in the interlamellar
space, while one fourth lie in the intralamellar space between the silica and alumina
sheets [Slade and Davies, 1991]. When kaolinite gets heated at temperatures above 700
K, sequential loss of hydroxyls has been observed [Slade et al., 1991]; those hydroxyls
Chapter 2. Background and literature review 14

situated at the interlamellar space being easier to get removed than the intralamellar
ones.

Kaolinite has got many interesting properties such as fine particle size, non-abrasiveness,
chemical stability and low viscosity at high solid content. Moreover, its structure exhibits
very little shrinkage and swelling [Grim, 1953]. Owing to these properties, it has got
several industrial applications notably as a filler in paper, rubber and paint industry
[Murray, 1963]. Furthermore, it can be utilized as supplementary cementitious material
in cement and concrete industry after being properly calcined [Sabir et al., 2001].

Figure 2.3: The arrangement of tetrahedral silica and octahedral alumina layers to
form kaolinite structure

In a comparative study to assess the reactivity of different clays, it turns out that thermal
activation of kaolinite displays high pozzolanic reactivity compared to other clays such as
montmorillonite and illite [Ambroise et al., 1985; Fernandez et al., 2011]. Consequently,
Kaolinite has been the interest of many researchers in the cement area, as it can be
thermally activated to produce the highly reactive pozzolanic material, metakaolinite.

2.3.1 Calcination of kaolinite

Thermal decomposition of kaolinite at moderate temperature (700-1000 K) yields amor-


phous structured material called metakaolinite (Al2 Si2 O7 ), a material that offers good
properties as supplementary cementitious material [He et al., 1994; Shvarzman et al.,
2003]. Thermal exposure beyond a definit point will result the formation of spinel-type
Chapter 2. Background and literature review 15

(Al4 Si3 O12 ) of phase along with amorphous silica, after which crystalline phases of mul-
lite (Al6 Si2 O13 ) and cristobalite (SiO2 ) form. The appearance of crystalline phases may
cause a decline in pozzolanic reactivity of the calcined material [Sabir et al., 2001].

A complete structural transformation of kaolinite mineral during thermal treatment


passes through a sequence of reactions [Brindley and Nakahira, 1958; Ptáček et al., 2011],
that occur at different temperatures. The whole transformation may be represented by
the following reaction scheme:

373K
H2 O(l,f ree) −−−→ H2 O(g) 4H = −2242(kJ/kg) (2.1)

700−1000K
Al2 Si2 O5 (OH)4 (Kaolinite) −−−−−−−→ Al2 Si2 O7 (M etakaolinite) + 2H2 O 4H = −632(kJ/kg) (2.2)
1000−1300K
2(Al2 Si2 O7 ) −−−−−−−−→ Al4 Si3 O12 (Spinel) + SiO2 4H = +230(kJ/kg) (2.3)
≥1300K
3(Al4 Si3 O12 ) −−−−−→ 2(Al6 Si2 O13 )(M ullite) + 5SiO2 4H = +84(kJ/kg) (2.4)
≥1473K
SiO2 (Amorphous silica) −−−−−→ SiO2 (Cristobalite) 4H = +42(kJ/kg) (2.5)

With regard to this study, the first two reactions, namely evaporation and dehydrox-
ylation are the most important reactions. Especially, the dehydroxylation reaction is
the one that needs to be optimized as the most pozzolanic material (metakaolinite) is
obtained at this stage. The remaining reactions that may occur at or beyond 1300 K are
phases transformations that deplete the amount of metakaolinite into spinel and mullite
phases; and hence need to be controlled.

Generally the above reactions are characterized by complex solid state reactions that
are influenced by the crystallinity of the kaolinite sample [Cabrera and Eddleston, 1983],
particle size [Lahiri, 1980], vapor pressure [Brindley et al., 1967] and heating rate. De-
spite the wealth of published works on the dehydroxylation of kaolinite, there is no
general consensus over the kinetics and mechanism of the entire reaction.

The calcination of such clay could be accomplished either by flash calcination or soak
calcination. Flash calcination is achieved by rapid heating (≈ several tenths of seconds)
followed by rapid cooling of the powdered clay material suspended in gas; whereas
soak calcination is achieved by slow heating for relatively long periods. According to
the literature [Salvador, 1995], flash-calcined products have shown peculiar structural
properties and reactivity that makes them a potential clinker substitute. Thus, a good
knowledge on the nature, composition and methods of calcining the clay material are
crucial to obtain a highly reactive pozzolanic product. For instance, the crystallinity,
the amount of metakaolinite and the particle size of the calcined clay strongly affects its
performance as SCM.
Chapter 2. Background and literature review 16

2.3.2 Kinetics of dehydroxylation

Dehydroxylation of kaolinite is accompanied by the loss of chemically bonded hydroxyls


and consequently, a structural disorder in the kaolinite framework causes the forma-
tion of amorphous phase, namely metakaolinite. This phenomenon is revealed by the
appearance of an endothermic peak between temperatures 800-1000 K. Analysis of the
differential thermal patterns of this peak is a common way of studying the reaction
kinetics of the clay material. Thermal analysis on the dehydroxylation of kaolinite has
been studied by using different techniques to examine the value of kinetic parameters,
specifically the activation energy (E) and the frequency factor (A).

Generally kinetic parameters for solid state reactions have been studied either by model-
fitting or model-free (isoconversional) methods. Model fitting methods yield single value
of activation energy which cannot account for the variation of activation energy due to
the complexity of the solid state reaction. The values obtained by these methods are
averages that do not reflect changes in the kinetics and mechanism with the temperature
and the extent of conversion. Whereas, model-free methods allow kinetic predictions to
be accomplished as a function of the extent of reaction. As such, the kinetic analysis is
carried out over a set of kinetic runs and gives better estimation of kinetic parameters.
Both methods have been practiced to determine the kinetic parameters of kaolinite
dehydroxylation. None of them is free of flaws, but model free methods got an upper
hand.

The use of empirical models (model-fitting methods) to study the dehydroxylation of


kaolinite has been carried out to examine the kinetics and reaction mechanism of the
dehydroxylation reaction [Brindley et al., 1967; Dion et al., 1998; Murray and White,
1955; Sharp et al., 1966]. Most of these studies attempt to solve if the reaction mechanism
follows first order or diffusion controlled mechanism, which has been a dispute over
the years. Yet, dehydroxylation of kaolinte remains controversial, where some authors
suggest the reaction obeys a first-order kinetic law [Allison, 1955; Dion et al., 1998;
Murray and White, 1955] whereas others suggest a diffusion model [Criado et al., 1984;
Horvath, 1985; Redfern S., 1987].

Kinetic parameters based on isoconversional methods are usually obtained using ther-
mogravimetric experiments. Among these methods, Kissinger’s method is one of the
most popular methods to calculate kinetic parameters for dehydroxylation of kaolinite
clays [Kissinger, 1956]. Accordingly, when a reaction occurs in differential thermal anal-
ysis, the thermal properties of the sample vary with heating rate and this variation is
manifested by deflection of temperature peaks. Analyzing the shift in peak temperature
at different heating rates during thermogravimetric experiment is the core concept to
Chapter 2. Background and literature review 17

determine the kinetic parameters. A complete derivation of the method can be referred
elsewhere in the literature [Chen et al., 1993; Kissinger, 1957, 1956; Llópiz et al., 1995].

Although the kinetics of kaolinite dehydroxylation has been broadly studied, there is
no single activation energy and frequency factor that characterize the dehydroxylation
process. Table 2.2 provides a summary of kinetic parameters of kaolinite dehydroxy-
lation from literature. The reason behind the diversity of values could be due to their
dependence on the natural composition, crystallinity and particle size distribution of
the kaolinite sample. The fact that dehydroxylation of kaolinite is a complex process
involving many individual steps, has been demonstrated by the dependence of activation
energy on the extent of conversion [Ortega et al., 2010].

Table 2.2: A summary of kinetic parameters of kaolinite dehydroxylation from liter-


ature

E(kJ/mol) A(s−1 ) Method Source Reference


177 4.57 × 108 Thermogravimetry Florida,USA Kissinger [1956]
162 1.26 × 107 Thermogravimetry Georgia,USA Kissinger [1956]
242 2.21 × 108 Effluent gas analysis Rep.Czech Ptáček et al. [2010]
196 9.6 × 108 TGA India Saikia et al. [2002]
195 8.58 × 1014 DTG Rep.Czech Ptáček et al. [2011]
193 1.70 × 107 DTA USA Bellotto et al. [1995]
163 2× 1012 DTA USA Levy and Hurst [1993]

2.4 Modeling thermal calcination of kaolinite

Today the use of computational modeling to tackle complex engineering problems in


the quest for optimal solutions in many fields of studies is a common practice. Hence,
mathematical models have been used to harvest any process related characteristics for
the calcination of kaolinite such as dehydroxylation mechanism, conversion and phase
transformation.

There has been little effort to understand calcination of kaolinite clays through modeling.
Indeed significant advances have been made in modeling fluid-solid, mainly gas-solid re-
actions [Bhatia, 1985; Georgakis et al., 1979; Patisson et al., 1998; Szekely and Propster,
1975; Wen and Wang, 1970]. Most of them are based on grain model and shrinking core
models to study the heat and mass transfer during reactions that involve gas-solid inter-
actions. However, these models consider several assumptions. Among them are steady
state and pseudo steady state approximations that ignore the accumulation term in the
gaseous phase [Bhatia, 1985; Szekely and Propster, 1975]. Such assumptions are mainly
Chapter 2. Background and literature review 18

made to simplify and circumvent mathematical difficulties; and are not recommended
in transient phenomena [Wen and Wang, 1970]. The isothermal assumptions rendered
in most grain and shrinking core models is another drawback of such models [Georgakis
et al., 1979]. When most of gas-solid reactions involve exothermic and endothermic re-
actions, it may impose temperature variation affecting the internal and external heat
transfer in the grain, thus, isothermal assumptions may only be used for a known cases
[Patisson et al., 1998].

In an attempt to study flash calcination of kaolinite clay, Salvador and Davies [1994]
presented a simplified thermochemical model which is used to examine the behavior of
kaolinite particles that are plunged into a hot gas atmosphere. In this model, different
particle sizes were investigated for their calcination behavior at different gas temper-
atures (823–1273 K). The model predicts fast conversion rates at higher temperatures
(1273 K). For instance, 100 µm particles are observed to dehydroxylate in 0.1 seconds.
The fast heating rate observed during flash calcination causes rapid generation of vapor
and is suggested to be the reason for particle decrepitation observed during flash dehy-
droxylation of kaolinite [Bridson et al., 1985]. However, this model does not consider the
effect of intra-particle processes that could influence dehydroxylation process. Favergeon
et al. [2013] also presented a kinetic model for kaolinite dehydroxylation at grain scale
that could be applied for heat and mass transfers at reactor scale. The kinetic model is
entirely based on the dehydroxylation reaction as a rate limiting reaction where several
built-in empirical models are used to monitor dehydroxylation reaction. The model is
able to predict the dependence of dehydroxyalation rate on calcination temperature and
vapor pressure. Moreover, the kinetic rate is found to be significantly affected by the
powder height in the reactor. This model is based on complex variables such as areic
frequency of nucleation, areic reactivity of growth, vapor pressure and so on that require
to apply several assumptions.

The above mentioned models by [Salvador and Davies, 1994] and [Favergeon et al., 2013],
however, have some limitations when addressing the complex behavior of kaolinite clay
calcination at different process conditions. For instance, none of them predicts the
composition of the calcined material and the phase transformation of kaolinite particle
when the temperature exceeds 1273 K.

On the other hand, modeling the calcination of limestone particles has been extensively
studied. Many of the investigations have been done on the calcination and simultaneous
sintering processes of limestone particles with the aid of mathematical models. Borg-
wardt [Borgwardt, 1985] found that the calcination reaction was kinetically controlled
except for the final stage of reaction where the diffusion of CO2 through the product layer
was rate limiting. Silcox et al. [Silcox et al., 1989] also developed a calcination model
Chapter 2. Background and literature review 19

of limestone (CaCO3 ) and (Ca(OH)2 ) particles where the decomposition of the parent
material at the reactant–product interface was described by a shrinking core model by
solving kinetic and transport equations for the gas and solid species. The model reliably
predicts the effects of particle size, temperature, time and relative rates of surface area
development. The study concluded that the escape of CO2 does not significantly slow
down the calcination reaction rate for small particles, and thus smaller lime particles
produce more reactive CaO particles in shorter period and at lower temperatures. But,
the hydrate particle produces more reactive CaO than do carbonates. This implies rapid
calcination rate for the hydrates, Ca(OH)2 than carbonates, CaCO3 .

The model developed by Hu and Scaroni [Hu and Scaroni, 1996] considers the heat and
mass transfer and chemical kinetics for calcination of pulverized limestone particles under
furnace injection conditions. The model show a significant influence of heat transfer,
mass transfer and chemical kinetics on the calcination rate of limestone. Moreover, due
to location-dependent calcination process, a gradient in temperature and CO2 partial
pressures are observed. Under such conditions, the calcination of smaller limestone
particles (≈ 63µm) is observed in less than 0.2 seconds at 1473 K furnace temperature.

Takkinen et al. [2012] also investigated the heat and mass transfer phenomena during
calcination of limestone through modeling. Two modeling approaches were used, namely
the shrinking core model and a transient numerical particle model where the mass,
momentum and energy equations are solved during calcination. The numerical particle
model depicts faster and uniform conversion for smaller particles at different calcination
stages. Furthermore, high temperature is observed to decrease the reaction time. Over
all, the importance of advection in the intraparticle transport during calcination was
highlighted for the numerical particle model. In their study the applicability of shrinking
core models towards lime calcination is compared with the numerical particle model.
Both models show certain degree of differences in their output, the difference being
significant for small particle sizes and higher CO2 concentrations in the surrounding gas.
These differences are however modest when the model assumes an apparent reaction rate
than infinite reaction rate.

Other calcination studies on limestone has also been demonstrated for the purpose of
cement production either at the precalciner or at the rotary kiln [Fidaros et al., 2007;
Mikulčić et al., 2012]. Their investigation focuses on the impact of partial pressure,
particle size distribution and porosity on the degrees of conversion. Alike the above
studies, the influence of temperature, particle size and partial pressure of CO2 on the
calcination process has been reported.

The usefulness of such limestone models in this study is the similarity of the calcination
process and thus, the modeling approach is somehow similar. However, the effect of
Chapter 2. Background and literature review 20

process variables during calcination of limestone is quite different from that of kaolinite.
i.e., the product of limestone calcination which is CaO may not further transform into
a different phase that affect the performance of cement. Hence, the severe calcination
condition and long residence time at the kiln system might have little effect on final
property of lime, if not on the cost of production. The exothermic nature of some
reactions involved during kaolinite rich clay calcination, however, makes the temperature
control more crucial and also more difficult, in order to obtain the optimum calcines.

2.5 Experimental study on flash calcination of kaolinite

Flash calcination of kaolinite has been investigated by several authors [Bridson et al.,
1985; Meinhold et al., 1992; Meinhold and Salvador, 1994; Slade and Davies, 1991;
Slade et al., 1991, 1992]. In most cases the powdered kaolinite clay is exposed to high
temperature(≈ 1273K) for short period of time, usually 0.5 to 1 second. Any change
in characteristics of the calcines is examined by rapid cooling at various stages during
calcination. Thus, flash calcined kaolinites undergo structural changes such as internal
voids. These changes are due to rapid generation of vapor, especially for residence times
beyond 0.3 seconds [Slade and Davies, 1991]. The flash calcines also display lower density
than the original kaoilinite sample [Bridson et al., 1985]. However, as the residence time
inside the calciner increases, the specific gravity progressively increases.

Flash calcination tests are also studied using different laboratory scale and pilot scale
flash calciners to examine the mechanism of dehydroxylation reaction [Meinhold and
Salvador, 1994]. These tests are carried out at different temperature and residence
time. In all cases, the rate of flash dehydroxylaton is influenced by diffusion of vapor
and the process is represented by 3 dimensional diffusion model. However, a change in
mechanism is observed at higher degrees of dehydroxylation. The different activation
energies calculated for the calcines produced by the different flash calciners may be
explicable due to the impact of vapor pressure inside the calciner that change the gas
flow characteristics, and also due to different sample introduction methods for each
calciner.

In a comparative study of calcined kaolinite particles from an industrial flash calciner and
conventional rotary kiln calciner, the chemical composition of the calcined metakaolinite
product is not influenced by the method of calcination [San Nicolas et al., 2013]; rather
the physical property of calcines is affected significantly. Since the calcination process
in rotary kiln calciner is achieved at low temperatures 920–973 K for about 3–5 hours,
agglomeration of the calcined product is observed. In an industrial flash calciner, the
process is accomplished in few tenths of seconds at temperatures 1273–1473 K. The
Chapter 2. Background and literature review 21

flash calcines display spherical morphology compared to the conventional kiln products,
furthermore, the flash calcines show high proportion of mullite due to direct contact
between the flame and clay particles inside the calciner.
Chapter 3

Single particle calcination model

This chapter describes the numerical effort used to model the calcination of kaolinite
rich clay particles. It briefly discusses the set of governing equations that characterize
the energy and fluid flow during calcination along with boundary conditions and method
of discretization used to solve the equations. It also gives an overview of the modeling
tools used in this study.

3.1 Fundamentals of CFD

Computational Fluid Dynamics (CFD) involves the analysis of a set of partial differen-
tial equations that characterize the flow of a fluid. For instance, consider the general
transport equation which is given as:

∂(ρψ)
+ ∇ · (ρUψ) = ∇ · (D∇ψ) + Sψ (3.1)
∂t

The above equation represents the flow and transport of fluids in which first two terms
in the left side of the equation correspond to the transient and convective terms, respec-
tively. The right side equations correspond to the diffusion and source terms, respec-
tively. Where ψ is a dependent variable transported due to the existence of a velocity
field, U and the gradient of the property transported, ∇ψ. The source term, Sψ accounts
for any sources or sinks that either create or destroy ψ.

A common CFD practice involves the discretization of the computational domain into
a number of cells or control volumes. The equation is then integrated over the control
volume so as to discretize the partial differential equations to algebraic equations by
using finite volume method. The set of algebraic equations are solved to calculate the
values at cell centers, after the initial and the boundary conditions of the problem are

23
Chapter 3. Single Particle Calcination Model 24

specified properly. In this way CFD provides an approximation to the analytical solution
of the general governing equations.

3.2 Description of the clay particle calcination model

The problem under investigation may be understood in such a way that when a kaolinite
clay particle initially at room temperature is exposed to a hot environment, different
physical and chemical processes may take place. Among them are heat transfer from the
surrounding, evaporation of any available moisture, dehydroxylation and transformation
of kaolinite into different phases sequentially. In order to substantiate the entire process
numerically, a particle model is developed.

In practice, the shape of such clay particles may be better represented by flat sheet
(flake-like). Due to the lack of accurate description of particle shape, the clay particles
are assumed as spheres, as most commonly used in particle shape approximation (e.g.,
pulverized coal particles), and an one-dimensional (1D) spherical particle model is de-
veloped to simulate thermal calcination of these particles. Figure 3.1 illustrates the 1D
calcination model showing the interaction of the clay particle with its surrounding. As
shown in the figure, the exchange of mass and heat with the surrounding gas is accom-
plished by radiation and convection through a gas film surrounding the particle, where
the temperature, mass concentration and other physical properties in the gas film are
evaluated using a simple 1/3rd rule [Abramzon and Sirignano, 1989].

As heating continues, the kaolinite clay particle conversion proceeds by releasing free
water and crystal water out of the particle to yield the amorphous material called
metakaolinite. When the particle temperature reaches certain level beyond complete
dehydroxylation, phase transformation of metakaolinite commences. Depending on the
process conditions, the particle conversion and transformation of the clay product is
examined based on the series of reactions given in section 2.3.1, reactions (2.1)–(2.5).
The sequence of reactions is governed by the temperature and kinetic parameters at
each reaction step; as such, high temperature and fast reaction rates may increase the
speed of phase transformation into unnecessary products such as mullite. Therefore,
the kinetic parameters for each reaction need to be carefully picked from literature or
experimentally determined.
Chapter 3. Single Particle Calcination Model 25


 T   T  T 
1
T
 film s 3  s
 1 
Y  Y  Y Y 
 j,film j,s 3  j, j,s 

Ambient Flow
T , Y j ,
Clay Particle

U(r), P(r) r=rs


r=0
T(r), Yj(r) Ts, Yj,s

Imaginary Gas Film around Clay


Particle, Tfilm, Yj,film

Figure 3.1: A sketch of 1D clay particle calcination model

3.2.1 Main assumptions

The main assumptions used in the model development are:

• The clay particle is spherical in shape and homogeneous in composition,

• 1D profile, i.e., the dominant variations in key parameters occur in the radial
direction (from the particle center to particle surface)

• Local thermal equilibrium exist within the clay particle, i.e., different phases at
the same temperature locally.

• Possible shape changes due to swelling or shrinking of the particle are neglected,

• The porosity of kaolinite particle hardly changes up on thermal treatment unless


it undergoes sintering beyond 1373 K [Chen et al., 2003]. Hence, constant particle
porosity is assumed,

• The heat and mass transfer conditions at the particle surface are symmetrical,

• Gas species and the gas mixtures follow the equation of state for an ideal gas,

• The same value of effective diffusivity (Deff ) has been assumed for all the gaseous
species, such assumption has been commonly used in literature [Lu et al., 2008;
Takkinen et al., 2012].
Chapter 3. Single Particle Calcination Model 26

3.2.2 Governing equations

Like any modeling process, numerical study of kaolinite calcination begins with a physical
model that is described by a set of PDEs which are developed based on conservation laws
of physics. The general governing equations that characterize the thermal conversion of
kaolinite during calcination are summarized as follows:

∂(ερg )
+ div(ερg U) = Sg (3.2)
∂t
η
U = − ∇P (3.3)
µ
∂(ερg Yj )
+div(ερg UYj ) = div(ερg Deff ∇Yj )+SYj (3.4)
∂t
∂ X  X 
ρi hi +ερg h+ρf w hf w +div(ερg Uh) = div(keff ∇T )+div hj ερg Deff ∇Yj +Sh
∂t
i j
(3.5)
∂(ρs,i Ys,i )
= Sy,i (3.6)
∂t

In the above equations, Eq. 3.2–3.4 represent the continuity equation, momentum equa-
tion modified with Darcy’s law in a porous medium, and the species transport equations
for the gaseous phases, respectively. Eq. 3.5 is a combined energy equation where the
influence of gases, solids and moisture on the transfer of energy is incorporated. The
last equation represented in Eq. 3.6 is the continuity equation for solid species.

The rate expression and kinetic parameters for the reactions involved during calcination
of kaolinite clay are represented as shown in Table 3.1.

Table 3.1: The expression of reaction rate and kinetic parameters used in the model

Reaction Rate expression A (s−1 ) E 4H Reference


(kJ/mol) (kJ/kg)
2.1 ṙ1 = ∂ρ∂tfw =k1 ρf w 5.13 × 1010 88 -2242 [Bryden and Hagge,
2003]
2.2 ṙ2 =∂ρkl /∂t=k2 ρkl 6.3 × 109 180 -632 (A,E):this work, 4H:
[Weber and Roy, 1965]

2.3 ṙ3 = ∂t (ρmk )=k3 ρmk 1.7 × 1016 405 230 [Gerardin and Sundare-
san, 1994]
∂ρsp
2.4 ṙ4 = ∂t =k4 ρsp 9.1 × 1015 424 84 (A,E): [Ptáček et al.,
2012], 4H: [Holm, 2001]
2.5 ṙ5 = ∂ρ∂tsil =k5 ρsil 1.75 × 1010 274 42 (A,E): [Ptáček et al.,
2012], 4H: [Holm et al.,
1967]
Chapter 3. Single Particle Calcination Model 27

The expression of key physical and transport properties used in the above governing
equations are summarized in Table 3.2.

Table 3.2: The Physical and transport properties and their expression in the model

Property Value/Expression Reference


P
Sensible enthalpy of gas mixture, h Y h
R Tj j j
Sensible enthalpy of species, hj Tref Cp,j (T )dT
Rate constant of k th reaction, kk Ak exp(−Ek /RT )
Effective conductivity, keff εkg + (1 − ε)ks [Takkinen et al., 2012;
Tavman, 1996]
Conductivity of solid, ks 0.3 [Michot et al., 2008]
ε 1 1 −1
Effective diffusivity, Deff + [Benitez, 2009; Silcox
τ DAB Dk
et al., 1989]
Molecular diffusivity, DAB −2.775 × 10−6 + [Bolz and Tuve, 1976]
4.479 × 10−8 T +
10−10 T 2
1.656 ×q
T
Knudsen diffusivity, Dk 48.50dp M Wg [Benitez, 2009; Silcox
et al., 1989]
Pore size, dp 0.3 × 10−6 [Diamond, 1970]
Permeability, η 1 × 10−12 [Reinecke and Sleep,
2002]
Particle emissivity, ω 0.7 [Bergman et al., 2011]
Tortuosity, τ 1.5 this work

The expression of the source/sink terms that appear in the above set of governing equa-
tions is based on the rate expression given in Table 3.1. Since evaporation and dehydrox-
ylation are the only reactions that generate gaseous species; the source/sink terms that
appear in the continuity equation, Sg and in the transport equation, SYj are basically
equivalent. The contribution of other gases (O2 and N2 ) during reaction is is negligible.
The source/sink term appearing in the energy equation is the heat effects summed over
all the five reactions. The expression of all source/sink terms are summarized as follows;

2M WH2 O
Sg = SYj = ṙ1 + ṙ2
M Wkl
k=5
X (3.7)
Sh = ṙk 4Hk
k=1

The source term expression for solid species, SYi is based on the rate of conversion of
individual species during calcination reaction and is summarized in Table 3.3.

Temperature dependent heat capacity of solid species is used in the study [Knovel, 2013;
Robie et al., 1979]. The heat capacity values for each solid components is summarized
in Table 3.4.
Chapter 3. Single Particle Calcination Model 28

Table 3.3: The expressions for solid species source/sink terms during calcination of
kaolinite clays

Source(sink) term Expression


For kaolinite, Skl −ṙ2
For metakaolinite, Smk ṙ2 (MWmk /MWkl ) − ṙ3
For spinel, Ssp ṙ3 (MWsp /2MWmk ) − ṙ4
For mullite, Smu ṙ4 (2MWmu /3MWsp )
For silica, Ssil ṙ3 (MWsil /2MWmk ) + ṙ4 (5MWsil /3MWsp ) − ṙ5
For cristobalite, Scr ṙ5 (MWsl /MWcr )

Table 3.4: Temperature dependent heat capacity of solid species

Heat capacity Value (J/(kg·K))


Cp,kl (200.58 − 0.795T + 4.42 × 10−3 T 2 − 8.25 × 10−6 T 3 + 7.65 ×
10−9 T 4 − 3.55 × 10−12 T 5 + 6.55 × 10−16 T 6 )1000/M Wkl
Cp,mk (146.5 − 0.51T + 2.98 × 10−3 T 2 − 5.68 × 10−6 T 3 + 5.29 ×
10−9 T 4 − 2.45 × 10−12 T 5 + 4.48 × 10−16 T 6 )1000/M Wmk

Cp,sp (146.5 − 0.51T + 2.98 × 10−3 T 2 − 5.68 × 10−6 T 3 + 5.29 ×
10−9 T 4 − 2.45 × 10−12 T 5 + 4.48 × 10−16 T 6 )1000/M Wsp
Cp,mu (7.55 × 102 − 2.94 × 10−2 T − 6.58 × 103 T −0.5 + 3.45 ×
10−6 T −2 )1000/M Wmu
Cp,sil (21.27 + 0.12T − 1.99 × 10−4 T 2 + 2.29 × 10−7 T 3 − 1.72 ×
10−10 T 4 + 7.295 × 10−14 T 5 − 1.29 × 10−17 T 6 )1000/M Wsil
Cp,cr (72.753T + 1.3 × 10−3 T + 4.13 × 106 T −2 )1000/M Wcr
† The heat capacity of spinel-type phase is not available in the literature. Thus, it is assumed to
have similar heat capacity as that of metakaolinite.

3.2.3 Boundary conditions

Boundary conditions at the particle center (at r = 0) are determined by symmetry as:

∂T ∂Yj ∂P
= 0; = 0; =0 (3.8)
∂t r=0 ∂t r=0 ∂t r=0

At the surface of the particle (r = RP ), the exchange of mass and heat with the sur-
rounding gas film is driven by the external convective mass transfer coefficient, hm and
convective heat transfer coefficient, hT , respectively; which in turn is calculated from
the empirical correlations available in the literature [Ranz and Marshall, 1952]. The Ste-
fan flow effect is not taken into account in this work, mainly due to the comparatively
small amount of water vapor flow released during clay calcination process. Thus, the
boundary conditions at the particle surface are:
Chapter 3. Single Particle Calcination Model 29


∂T 4
keff = hT (T∞ − Ts ) + ωσ(Trad − Ts4 )
∂t r=Rp
hT D p 1 1
Nu = = 2 + 0.64(Re) 2 (P r) 3
kg

∂Yj (3.9)
Dj,m = hm (Yj,∞ − Yj,s )
∂t r=Rp
hm Dp 1 1
Sh = = 2 + 0.64(Re) 2 (Sc) 3
DAB
P |r=Rp = Patm

The physical properties in the gas film around the particle are evaluated based on the
reference conditions that can be calculated by a simple 1/3rd rule [Abramzon and Sirig-
nano, 1989], from which the non-dimensional constants such as N u, P r, Re, Sc, and Sh
numbers are computed. Thus, the reference conditions in temperature and mass fraction
at the gas film are calculated as,

1 
Tf ilm = Ts + T∞ − Ts
3 (3.10)
1 
Yj,f ilm = Yj,s + Yj,∞ − Yj,s
3

3.2.4 Initial conditions

Apart from boundary conditions, the transient calcination model require well defined
initial conditions where initial values of flow variables are specified in the flow domain.
These conditions usually describe the states of the solid and gas phases at the beginning
of the process (t = 0).
The initial conditions acquired are:

T |t=0 = 298.15K; P |t=0 = Patm


(3.11)
YH2 O t=0 = 0; YO2 t=0 = 0.23; YN2 t=0 = 0.77

3.3 Discretization

Discretization of the calculation domain is achieved by systematically dividing the space


and dependent variables so that the governing differential equations could be expressed
with simple algebraic equations that can be solved easily. The discretization of the
governing equations in the present work is based on finite volume method (FVM), which
is one of the well studied schemes [Patankar, 1980; Versteeg and Malalasekera, 2007].
Chapter 3. Single Particle Calcination Model 30

The discretized equations are then assembled to a standard form that can be solved by
using TDMA (Tri-Diagonal Matrix Algorithm).

aP ψP = aW ψW + aE ψE + Su (3.12)

where ψ is a property that can be computed, for example the temperature, mass fraction,
pressure and so on. The subscripts P , W and E denote the position at current cell-
center, at the west and the east neighboring cell-centers, respectively. Su is the source
term, if any.

Since the transient particle model output is a function of time and position, the dis-
cretization is carried out for both of the domains. Thus, numerical discretization of
such time-dependent partial differential equations (PDE) need to be fully discretized in
both time and space. The governing equations are discretized in space first to transform
PDEs into ordinary differential equations (ODEs), followed by time integration of ODEs
to advance in time.

3.3.1 Spatial discretization

The discretization of the single particle model is accomplished by dividing it into a


finite number of control volumes (CV). In order to integrate the PDE’s over each cell,
the basic mesh is defined by a set of points (nodes) from the domain so that to each node
a control volume is assigned. The computational domain in this work is shown in Figure
3.2, where the spherical particle is divided into N equidistant nodes, 4r. To simplify
the notation in accordance with the standard notation widely used in many of the CFD
books, the letters P , W and E denote the nodes at ri , ri − 1 and ri + 1, respectively.
At these nodes, the scalar variables such as the temperature, pressure, density and mass
fraction are evaluated. Each control volume is bounded by faces that are located midway
between the grid points. These faces are represented by w and e which are located at
ri − 21 4r and ri + 12 4r, respectively. At these faces, velocity components are calculated.
In this study, discretization is based based on the upwind scheme for the convection
term, and the central difference scheme for the diffusion term.

3.3.2 Temporal discretization

Temporal discretization has been carried out using fully implicit method. The advantage
of this method is its unconditional boundedness and robustness for any size of time step
[Patankar, 1980; Versteeg and Malalasekera, 2007]. Hence, the governing equations are
Chapter 3. Single Particle Calcination Model 31

Clay particle

Control volume

W w P e E
i-1 i i+1 r
Center
∆r

Figure 3.2: Discretization of the one dimensional clay particle

temporally discretized over a time step (4t). A general expression for the transient term
in the PDEs during time marching of a scalar variable φ is given as;

∂ψ
= F (ψ) (3.13)
∂t

where F is anonymous function that may have any spatial discretization. The first-order
accurate implicit temporal discretization is given by ;

ψ − ψo
= F (ψ) (3.14)
4t

where the variable with superscript, ψ o denotes the value of the quantity at the previous
time, t − 4t, while the values of ψ are evaluated at current time, t. This way, the fully
implicit scheme is implemented to all transient governing equations.

3.3.3 Final discretized equations

As briefly discussed in the previous section, the final form of the energy, transport
and continuity equations are discretized and rearranged into a standard form aP ψP =
aW ψW + aE ψE + Su . The final solution is summarized by sorting the value of the
coefficients. Solving the discretized energy equation by using TDMA solver gives the
value of temperature. The value of coefficients for the discretized energy equation are
summarized in Table 3.5.
Chapter 3. Single Particle Calcination Model 32

Table 3.5: A solution to the discretized energy equation

Coefficient Solution of energy equation


aW (cell center) 0.0
   
keff A ερg Cp,g UA
aW (internal cells) 4r + 2
 w  w
keff A ερg Cp,g UA
aW (cell surface) +
 4r w  2
 w
keff A ερg Cp,g UA
aE (cell center) -
 4r e  2
e
keff A ερg Cp,g UA
aE (internal cells) 4r - 2
e e
aE (cell surface) 0.0
ρCp 4V
aP (cell center) 4t + aE + aW + (ερg Cp,g UA)e - (ερg Cp,g UA)w
ρCp 4V
aP (internal cells) 4t + aE + aW + (ερg Cp,g UA)e - (ερg Cp,g UA)w
ρCp 4V
aP (cell surface) 4t + aE + aW + (ερg Cp,g UA)e − (ερg Cp,g UA)w + Asurf hT + 4Asurf ωσ(TPo )3
(ρCp )o Tpo 4V P5
Su (cell center) 4t + k=1 ṙk 4Hk 4V
(ρCp )o Tpo 4V P5
Su (internal cells) 4t + k=1 ṙk 4Hk 4V
(ρCp )o Tpo 4V P5 4 + 3(T o )4 )
Su (cell surface) 4t + k=1 ṙk 4Hk 4V + Asurf hT T∞ + Asurf ωσ(Trad P

• The superscript, o , indicates the value of the parameter at the previous time step
P P
• ρCp = i ρi Cp,i + j ερg Yj Cp,j + ρf w Cp,f w

At the outer most control volume, the final discretized energy equation need to address
the heat source term Sh at the particle surface due to convection and radiation. Thus,
the heat source term Sh is integrated at the control volume to yield:
Z Z
4
Sh dV dt = hT Asurf (T∞ − Ts ) + Asurf ωσ(Trad − Ts4 ) (3.15)
4t cv

The nonlinear radiation source term is handled as described elsewhere in the literature
[Yin et al., 2010], where the linearized radiation term will have the form,
 
4
Asurf ωσ(Trad − Ts4 ) ≈ Asurf ωσ Trad
4
− (TPo )4 − 4(TPo )3 (TP − TPo ) (3.16)

Discretizing and integrating the transport equation solves for the mass fraction of gas
species, Yj . The solution of the transport equation sorted in accordance with the stan-
dard form of notation is summarized in Table 3.6.

Solving the continuity equation requires substituting the physical velocity at the east
and west interfaces using Darcy’s flow equation Eq. 3.3. The solution of the resulting
discretized equation solves for the pressure correction term, P0 as shown in Table 3.7.

The velocity at the outer most control volume is derived from the discretized continuity
equation as;
h S 4V
g (ρg − ρog )4V i
Ue = − + (ρg UA)w /(ρg A)e (3.17)
ε 4t
Chapter 3. Single Particle Calcination Model 33

Table 3.6: A solution to the discretized transport equation of gases

Coefficient Solution of transport equation

aW (cell center) 0.0


   
ερg Deff A ερg UA
aW (internal cell) 4r + 2
 w  w
ερg Deff A ερg UA
aW (cell surface) 4r + 2
 w   w
ερg Deff A ερg UA
aE (cell center) 4r − 2
 e  e
ερg Deff A ερg UA
aE (internal cells) 4r − 2
e e
aE (cell surface) 0.0
ερg 4V
aP (cell center) 4t + aE + aW + (ερg UA)e − (ερg UA)w
ερg 4V
aP (internal cells) 4t + aE + aW + (ερg UA)e − (ερg UA)w
ερg 4V
aP (cell surface) 4t + aE + aW + (ερg UA)e − (ερg UA)w + ρg Asurf hm
ερog Yj,p
o 4V
Su (cell center) 4t + Sy,j 4V
ερog Yj,p
o 4V
Su (internal cells) 4t + Sy,j 4V
ερog Yj,p
o 4V
Su (cell surface) 4t + Sy,j 4V + ρg Asurf hm Yj,∞

Table 3.7: A solution to the discretized pressure-correction equation.

Coefficient Solution of the pressure-correction equation.


η
aW µ4r (ρg A)w
η
aE µ4r (ρg A)e

aP aE + aW
ρg 4V (ρg −ρ0g )4V
Su ε − 4t + (ρg AU∗ )w − (ρg AU∗ )e

Here it must be noted that the correction pressure P0 is defined as a difference between
correct pressure field, P and the guessed pressure field, (P∗ ). The same is true for the
velocity field, as shown below,

P = P∗ + P0
η (3.18)
U = U∗ + (P0 − P0i+1 )
µ4r i

The indexes i and i + 1 denote the current control volume and its east neighbor control
volume, respectively.
Chapter 3. Single Particle Calcination Model 34

3.4 Iterative procedure

Once the appropriate equations, the boundary and initial conditions are set; the solution
methods involve an iterative scheme to arrive at simulation results. Since the system of
PDE’s is nonlinear, several iterations must be carried out during each time step of the
numerical solution until convergence is reached.

During the iterative convergence, some criteria may need to be fulfilled and it is usually
defined by acceptable error in some parameter values. It is also important to examine
whether the final time has been reached with proper convergence at each time step.
Figure 3.3 shows the iterative procedure used to obtain numerical solution for the particle
model.

Figure 3.3: Overall algorithm for the sequence of operations in the problem
Chapter 3. Single Particle Calcination Model 35

3.5 Modeling tools: C++ and gPROMS ModelBuilder

For simulating the calcination process of kaolinite clay particle, a stand alone C++
code has been successfully developed; the same process has also been simulated in a
commercial software called gPROMS.

• The model developed in c++ for single clay particle calcination is not only to
achieve a better understanding of the conversion process of a single clay particle
when it is suddenly exposed to a known calcination condition, but also to be inte-
grated, after proper reformulation, into advanced CFD simulations of the reacting
particulate flow system in the calciner later. CFD simulation of the entire calciner
is crucial in order to really understand its performance and ultimately to come up
with innovative conceptual design of smart and energy-efficient calciners.

• The ultimate purpose of using gPROMS here is for the calcination plant perfor-
mance evaluation and optimization by modeling the calciner and integrating the
calciner model into the whole system in the plant. Even though such a plan was
changed by the project consortium during the project, the comparison between
the standalone c++ code and commercial gPROMS can serve as a kind of model
cross-validation.

gPROMS is a commercial software abbreviated for general PROcess Modeling System.


It is a bundle of software tools with a common solver kernel to compute numerical so-
lutions and optimization problems [Pantelides and Barton, 1993]. These are created in
a central graphical user interface (GUI) which is called ModelBuilder. This tool can
also be used to start certain tasks, analyze and view results. Modeling and Simulating
in gPROMS are done by setting up different objects called entities with dedicated pur-
poses [Oh and Pantelides, 1996; Pantelides and Barton, 1993; PSE, 2013]. For instance,
gPROMS ModelBuilder requires to have at least the VARIABLE TYPE, MODEL and
PROCESS entities. The upper and lower bounds of a variable are declared under the
entity VARIABLE TYPE. The set of equations, assignments, parameters and initial
conditions together with event conditions are specified under the entity MODEL. Sev-
eral models can also be written and coupled to each other without worrying about
their hierarchy. The PROCESS entity is the main task that controls simulation activi-
ties, furthermore, this entity includes assignment of parameters, specification of initial
conditions and setting solver parameters. Figure 3.4 illustrates a screenshot of a well
posed calcination model and gPROMS ModelBuilder interface along with visualization
of output results.
Chapter 3. Single Particle Calcination Model 36

gPROMS uses a number of state-of-the-art hierarchical solvers. Among these are, DA-
SOLV for Differential-Algebraic Equations, BDNLSOL for Nonlinear Equations and the
MA28 and MA48 for sparse linear system of equations.

Although this study is not intended to compare the two modeling tools, the major
differences in using C++ code and gPROMS to simulate the calcination process are
highlighted as follows;

• While C++ needs to discretize the geometry and the governing equations manually,
gPROMS handles it by simply specifying the method of discretization and its order
of accuracy.

• C++ needs to write the final discretized equations in a standard form [Patankar,
1980; Versteeg and Malalasekera, 2007] so that it would be convenient to solve by
Tri-diagonal matrix algorithm (TDMA) or Thomas algorithm; whereas gPROMS
needs to write the equations as they appear in paper [PSE, 2013] and the mathe-
matics is handled by built-in solvers.

• While convergence and computational stability needs much effort in C++; initial-
ization and setting well posed system is the most difficult task in gPROMS.

Figure 3.4: A screenshot of gPROMS interface


Chapter 4

Experimental study

The experiments under this section not only provide a detailed data on the initial compo-
sition and property of the kaolinite rich clay material, but also offer an in-depth under-
standing on the thermal calcination of kaolinite rich clay sample into different products
under different calcination conditions. This chapter, therefore, is dedicated to the exper-
imental study performed to determine kinetic parameters, characterize the composition,
specific density, PSD and other physical properties of the clay material under investiga-
tion. The outcome from the experiments is used as an input for modeling study or model
validation.

4.1 Study of kinetic parameters

The kinetic parameters (A and E) of dehydroxylation of kaolinite rich clay particles are
determined by using the well known Kissinger’s approach [Kissinger, 1957, 1956], using
thermogravimetric experiments. The method is based on the shift in peak temperature
(the temperature at which the reaction rate is maximum), when a reaction occurs at
different heating rates. The shift in peak temperature is clearly shown in in Figure 4.1
when DTA experiments are performed at different heating rates for kaolinite rich clay
sample.

37
Chapter 4. Experimental study 38

Figure 4.1: The shift in temperature during dehydroxylation of kaolinite rich clay at
different heating rates, θ (◦ C/min)

In order to summarize the derivation of Kissinger’s equation briefly, we begin with the
general expression of decomposition reaction as,


= kf (α) (4.1)
dt

where α is the the degree of dehydroxylation, k is the rate constant, and f (α) is a kinetic
model that dictates the behavior of the dehydroxylation reaction.

According to Kissinger, the the kinetic model is represented by order based kinetics,
where f (α) = (1 − α)n . Hence, the above equation is rearranged into:


= k(1 − α)n (4.2)
dt

where n is the empirical kinetic exponent of the reaction. Substituting the rate constant,
k with Arrhenius relationship and rearranging and integrating the equation gives the
final Kissinger’s expression as shown in Eq. 4.3. Further information on the derivation
of the method can be referred in the literature [Chen et al., 1993; Kissinger, 1957, 1956;
Llópiz et al., 1995; Ptáček et al., 2011].
   
θ E AR
ln =− + ln (4.3)
Tp2 RTp E

where, θ is the heating rate (K/min) and Tp is the maximum reaction rate temperature
(K). The plot ln(θ/Tp2 ) versus 1/Tp is a straight line whose slope is −E/R and an
Chapter 4. Experimental study 39

intercept ln(AR/E). Although the method has been criticized and modified by several
authors [Baumann et al., 2010; Boswell, 1980; Sesták et al., 2014], it still remains popular
to determine kinetic parameters of decomposition reactions and other solid-gas reactions.

The kinetic exponent or Avrami’s constant (n) has also been calculated based on the
characteristics of the DTG curve [Augis and Bennett, 1978], where its value is approxi-
mated as:
2.5RTp2
n= (4.4)
W1/2 E
The reaction mechanism has been interpreted based on the value of Avrami’s constant,
as it is commonly practiced in literature [Criado et al., 1984; Hankock and Sharp, 1972;
Ptáček et al., 2011; Saikia et al., 2002].

4.1.1 Thermogravimetric study

Dehydroxylation of kaolinite clay is carried out by using thermogravimetry method. It


simultaneously measures both the heat flow (DSC) and weight changes (TGA) associated
with transitions in a material as a function of temperature and time in a controlled
atmosphere. The general procedure of the test is described as follows. Approximately
about 20 mg (±2) of clay sample is placed in a Pt crucible. The sample is heated to
a maximum temperature of 1373 K at different heating rates (5, 10, 20, 30, 40 and
50 K/min) in nitrogen environment at flow rate of 100 cm3 /min. Relevant data are
collected, in which the weight loss from TGA and its derivative DTG are plotted against
temperature, as shown in Figure 4.2. Moreover, the DTG data were analyzed to extract
the peak temperature (Tp ) and the half peak width (W1/2 ).

Figure 4.2: A representative thermogram for an experiment done at a heating rate


of 20 K/min
Chapter 4. Experimental study 40

Two TGA experiments are carried out utilizing different TGA models. The first experi-
ment is done using model SDT 2960 and the second is STA 409 PC, which are illustrated
in Figure 4.3. The same procedure is used for both experiments except that the sample
size is well controlled for each test in the second experiment.

Figure 4.3: TGA equipment used during study

The DTG curve for each test is examined to derive useful information such as the peak
temperature, the half width maxima (W1/2 ) and the values of kinetic exponent (n). The
Experimental peak characteristic values obtained from DTG curve at different heating
rates are shown in Table 4.1 and 4.2.
Table 4.2: Experimental peak char-
Table 4.1: Experimental peak char-
acteristic values from Experiment 2.
acteristic values from Experiment 1.
θ (K/min) Tp (K) W1/2 n
θ (K/min) Tp (K) W1/2 n
5 778 66 1.28
5 769.6 64 1.1
10 795.5 70.1 1.24
10 785 69 1.04
20 816.4 72.3 1.22
20 805 75.1 1.03
30 825 74.2 1.21
30 815.32 78.5 1.02
40 832.15 80.5 1.12
40 828.6 85.2 1.0
50 837.6 81.7 1.12

Based on experimental results collected for TGA experiments, the plot ln(θ/Tp2 ) versus
1/Tp is illustrated in Figure 4.4, where the activation energy is calculated from the slope.
Chapter 4. Experimental study 41

Figure 4.4: Plot of ln(θ/Tp2 ) versus 1/Tp to determine kinetic parameters

4.2 Gas Suspension Calciner (GSC)

The calcination test is performed in a gas suspension calciner located in FLSmidth R& D
Center Dania (Denmark). The schematic sketch of the GSC is illustrated in Figure 4.5.
The main objective of this test is to examine the property of calcines after it gets exposed
to various calcination temperatures. For this purpose, a dried and crushed kaolinite rich
clay sample is used as a feed material with initial composition of 90-95 % kaolinite and
5-10 % quartz. The feed has particle size distribution as shown in Figure 4.6.

The feed material is introduced into the system at a feeding rate of 6 kg/hr. The hot
gas generated in the main burner is drawn into the calciner from the bottom and pulled
through the reactor tube together with the feed particles by an extraction fan. The hot
gas/particle flow is mixed with a stream of quenching air at the exit of the GSC, and
finally enters the filter to separate the product from the gas flow.

The system has provision of secondary burners that may be put into operation to obtain
a uniform temperature profile in the calciner. Hence, two sets of experiments are per-
formed. The first sets of experiments are accomplished without support burners where
calcination tests under 1073, 1173, 1273 and 1373 K are performed. During this test a
temperature drop was noticed along the calciner. The temperature across the calciner is
measured with the help of thermocouples that are put at four different locations along
the GSC. The second set of experiments are performed by using support secondary
burners to maintain constant calcination temperature at 1073 and 1273 K all over the
calciner.
In all of the six tests made, the clay particles are expected to have about 0.5 s residence
Chapter 4. Experimental study 42

Figure 4.5: A schematic sketch of the gas suspension calciner along with the position
of secondary burners (SB) and thermocouples (TC).

time inside the calciner. The calcined clay material is then collected after each test
for further analysis. Table 4.3 presents the calcination conditions for the experiments
carried out in GSC and detailed description of each experimental setup and sample ID.
Those experiments carried out when secondary burners are not operational (EXP-I) ex-
hibit a significant temperature drop along the calciner. When secondary burners are
operational (EXP-II), the temperature inside the GSC is observed to be stable.

Table 4.3: Calcination conditions of the GSC and details of sample ID

Calcination temperature (K) Secondary Calcined Temperature drop (K)


Test
Nominal Average burners product ID TC-1 TC-2 TC-3 TC-4
Exp-I 1073 964 NO MK964NB 1067 979 921 893
1173 1058 NO MK1058NB 1168 1078 1022 972
1273 1126 NO MK1126NB 1269 1158 1062 1012
1373 1214 NO MK1214NB 1362 1244 1147 1081
Exp-II 1073 1073 Yes MK1073WB 1075 1070 1062 1061
1273 1273 Yes MK1273WB 1271 1276 1263 1264

4.3 Characterization test of the raw feed and the calcines

Today, the significant advance in X-ray diffraction (XRD) and X-ray fluorescence (XRF)
techniques enabled to fully characterize clay minerals [Bish and Post, 1993; Chung and
Chapter 4. Experimental study 43

Smith, 1999; Snellings et al., 2010]. However, the complex nature of clays make char-
acterization difficult and thus, it needs to be supplemented with other methods for
accurate quantification. This section presents different characterization tests that have
been made to fully characterize both the feed and the calcined clay material.

4.3.1 Mineralogical composition and PSD

The mineralogy of the clay material under investigation is determined by a combination


of results from chemical analyses by X-ray fluorescence and X-ray diffraction on both
oriented and random powder samples, supplemented with other thermal methods such
as thermogravimetry.
The kaolinite rich clay sample has chemical composition as shown in Table 4.4. This clay
sample is pretreated and conditioned prior to using it as feed material. The treatments
are usually physical and doesn’t affect the composition of the clay sample.

Table 4.4: Chemical composition of kaolinite rich clay sample

Chemical composition, wt.%


SiO2 Al2 O3 Fe2 O3 CaO TiO2 P2 O5 K2 O Na2 O LOI
47.9 35.06 0.77 0.2 1.29 0.11 0.12 0.06 13.65

The particle size distribution of the clay sample is measured using Malvern equipment
with laser diffraction by measuring the angular variation in intensity of light scattered
when a laser beam passes through a dispersed clay sample. It is based on the principle
that particles passing through a laser beam will scatter light at an angle that is directly
related to their size: large particles scatter at low angles, whereas small particles scatter
at high angles. Thus, a collection of particles will produce a pattern of scattered light
defined by intensity and angle that can be transformed into a particle size distribution.
The PSD of the feed and calcined clay is illustrated in Figure 4.6.

4.3.2 Density and surface area

The density of the feed clay material and the calcined clay material has been measured
by Gas Pycnometer Micromeritics, AccuPyc II 1340. The measurement and principles
are based on gas displacement method to measure volume accurately. An inert gas,
helium, is used as the displacement medium. The clay sample is sealed in the instrument
compartment of known volume, the appropriate inert gas is admitted, and then expanded
into another internal chamber. The pressure difference before and after expansion helps
Chapter 4. Experimental study 44

Figure 4.6: Particle size distribution of the feed and calcined clay materials.

to compute the sample volume. The density is then calculated by dividing the sample
weight by its volume.

Surface area of the feed and calcined clay material is measured using BET method which
is based on the adsorption of gas on a surface. The amount of gas adsorbed at a given
pressure allows to determine the total specific surface area in m2 /g.

4.3.3 Degree of dehydroxylation

The degree of dehydroxylation of the clay sample is studied by using LECO RC612
multiphase carbon and hydrogen/moisture analyzer shown in Figure 4.7. The equip-
ment features a state-of-the-art furnace control system, allowing the temperature of the
furnace to be programmed from near-ambient to 1373 K. The main purpose of this ex-
periment is to examine the degree of conversion or dehydroxylation by monitoring the
amount of crystal water in the clay sample.

The principle is based on examining the mass of water left after the clay sample has
been heated to high temperature inside the equipment. For such purposes, 250 mg of
the feed clay is placed into a system stabilized with inert nitrogen atmosphere. The raw
feed clay sample is held at 423 K for 15 minutes to make sure that the free water is
removed. Then, the clay sample is exposed to a heating rate of 120 K/min until 1273 K
and held at this temperature for about 7 minutes to secure a complete dehydroxylation
(i.e., a complete removal of the crystal water). A built-in Infrared detection method is
Chapter 4. Experimental study 45

used to quantify the amount of vapor as a weight percentage. The same procedure is
repeated for calcined samples. The final output from the instrument gives the amount
of water vapor in weight percentage, as shown in Figure 4.8.

Figure 4.7: LECO RC612 Multiphase Carbon and Hydrogen/moisture analyzer.

Figure 4.8: The amount of vapor in terms of weight fraction, as it is shown in LECO
RC612: (a) Feed (b) calcined clay at 1273 K (MK1273WB )

Finally, the degree of dehydroxylation, α is determined by comparing the crystal water


left in each calcined sample (mCS ) to the initial amount of crystal water in the raw feed
(mRF ), as shown in Eq. 4.5. When all crystal water is removed from the calcined sample,
the amount of water left (mCS ) approaches to zero and hence degree of dehydroxylation,
Chapter 4. Experimental study 46

α becomes closer to 1.

mCS
α=1− (4.5)
mRF
A summary of the vapor signals for all clay samples is illustrated in Figure 4.9, where
the intensity of water vapor evolved from the feed and calcined clays are plotted against
temperature. The observed decreasing trend in the intensity of vapor is due to the fact
that the calcination process in the GSC has driven out most of the crystal water from
the clay material.

Figure 4.9: A graphical plot of the intensity of vapor during LECO RC612 thermal
analysis for the feed and calcined clay products.

4.3.4 Composition of phases

X-ray diffraction is by far the commonest method used to determine the amount of
phases in clay sample. In this study, XRD is used to determine the qualitative and
quantitative phase analysis of clays. The analysis of the as-received and calcined sam-
ples is carried out using PANalytical CubiX PRO X-ray diffractometer as illustrated in
Figure 4.10. Rietveld analysis of randomly oriented clay powders is used to quantify
the composition of the calcined clay material after each run at its respective calcination
temperature. Quantification of phases is based on the comparison of the mass fraction
of each component with a known standard [Bish and Post, 1993], where 10 % anatase
(TiO2 ) is used as internal standard.
Chapter 4. Experimental study 47

Figure 4.10: The PANalytical CubiX PRO X-ray diffractometer used in this study.

Since X-ray diffraction determine the relative amount of each phase by comparing the
positions and intensities of the diffraction peaks against a library of known crystalline
materials, having known the amount of crystalline phases in the calcined material, the
amorphous material (metakaolinite) can be approximated indirectly as:
X
Yam = 1 − Yi,c (4.6)
i

where Yam is mass fraction of amorphous material and Yi,c is the mass fraction of crys-
talline phases in the calcined clay material. The approximated mass fraction of the
amorphous material may not entirely be metakaolinite, some unidentified none crys-
talline materials may share some part; for instance amorphous silica. It is known that
the clay material under study is rich in kaolinite with some quartz impurities. Since
kaolinite, quartz and mullite phases are crystalline they can be easily quantified with
XRD as shown in Figure 4.11. Based on this quantification, the amount of amorphous
phase (metakaolinite) is reasonably approximated.
Chapter 4. Experimental study 48

Figure 4.11: XRD pattern of the as received clay sample and its calcines at different
calcination temperature.
Chapter 5

Result and discussion

This chapter presents the outcome of the study supplemented with discussions in relation
to the available literature. Results are presented in different sections where the values
of kinetic parameters are computed, comparison of experimental data with model predic-
tion are established and composition of phases are explained. Lastly, a detailed results
predicted by the model at different calcination conditions are presented, with brief sen-
sitivity analysis on the impact of the most important calcination conditions such as gas
temperature, residence time and kinetic parameters.

5.1 Kinetic parameters

Based on the experiments performed in chapter 4, kinetic parameters for kaolinite de-
hydroxylation reaction are determined. The calculated experimental values of kinetic
parameters are given in Table 5.1 along with the “average” value which is explained
in Paper 2. Other kinetic parameters collected from literature are also given in the
same table for comparison purposes. Generally, the values are so diverse that they can
influence the rate of reaction significantly. Figure 5.1 illustrates the impact of kinetic
parameters on the rate constant, k, which is expressed in terms of Arrhenius relationship
as, k = Ai exp(−Ei /RT ).

49
Chapter 5. Result and Discussion 50

Table 5.1: A summary of kinetic parameters for kaolinite dehydroxylation from liter-
ature and experimental values of this study

E(kJ/mol) A(s− 1) Method Source Reference


177 4.57 × 108 Thermal Analysis, DTA Florida, USA [Kissinger, 1956]
193 1.0 × 109 Thermal Analysis, TGA Georgia, USA [Bellotto et al., 1995]
196 9.6 × 108 Thermal Analysis, TGA India [Saikia et al., 2002]
195 8.58 × 1014 Thermal Analysis, DTG Rep. Czech [Ptáček et al., 2011]
163 2.0 × 1012 Thermal Analysis, TGA USA [Levy and Hurst, 1993]
176 1.66 × 1011 TGA, SDT 2960 EU Experiment-I
189 2.7 × 109 TGA, STA 409 PC EU Experiment-II
180 6.3 × 109 – EU Average value

Kissinger Belloto et al. Saika et al. Ptacek et al.


Levy & Hurst Experiment‐I Experiment‐II "Average"
150

120
Rate constant, k[s‐1]

90

60

30

0
500 700 900 1100 1300 1500
Temperature, T[K]

Figure 5.1: The influence of kinetic parameters on the rate constant

Although the experiment is aimed at estimating the kinetic parameters quickly, differ-
ent values of activation energy and frequency factor are obtained for each experiment,
but within the range of values given in literature. A possible reason for the variation
could be the interdependency among the different factors considered. The sample mass,
sample particle size and packing [Schilling, 1990] are among the factors that may af-
fect the shape, precision and accuracy of the experimental results in thermogravimetry.
The sample mass might cause sample temperature to deviate from a linear temperature
change affecting the shape of thermogram during experiment. The particle size may
also cause a change in the diffusion of the evolved gases during reaction which alters the
reaction rate and hence the shape of the DTG curve [Stoch, 1984; Stoch and Waclawska,
1981]. Apart from the above mentioned reasons, the different values of kinetic param-
eters obtained above could also be due to the uncertainty associated with the method
itself, which has been under debate in literature [Michèle et al., 2011; Opfermann and
Flammersheim, 2003; Pijolat et al., 2005; Sesták et al., 2014].
Chapter 5. Result and Discussion 51

5.2 Experimental and modeling results of flash calcination


in the GSC: A comparison

As a first effort to validate the model, TGA experiments are used to compare the con-
version profile with model results. The details can be seen in Paper 1, where further
explanation on the little discrepancy observed at higher degree of conversion are made.

The other major effort to validate the model results is using experimental results from
the pilot scale GSC. Paper 2 describes in detail the comparison between the GSC ex-
perimental data and model predictions.

The process conditions inside the GSC are taken into account in the model. There
are two parameters that need attention: the particle size and calcination temperature.
The implementation of the real particle size and real temperature during calcination is
essential to fairly generate representative results. The particle size is implemented based
on the PSD of the feed material (see Figure 4.6), from which 6 representative particle
sizes are finally used (with proper volume percentages) in the model. The calcination
temperature is implemented based on the temperatures measured at different position
in the GSC. Thus, a piecewise-linear temperature profile is implemented in the model.
Detailed description of such implementation can be found in Paper 2.

For those tests done in the presence of secondary burners (Exp-II), the temperature pro-
file along the GSC is nearly constant and uniform. As a result, the nominal calcination
temperature is implemented directly in the model.

Under the above calcination conditions and proper implementation of temperature and
PSD, model results are compared with experimental data. The density and degree of
dehydroxylation are good indicators to evaluate structural changes during the course of
calcination [Davies, 1986; Slade and Davies, 1991]. Hence, a comparison in density is
illustrated in Figure 5.2.

As shown in Figure 5.2, when kaolinite rich clay particles are flash calcined at a given
calcination temperature for about 0.5 seconds, it displays a drop in density of the cal-
cined material at lower calcination temperatures (MK964NB and MK1058NB ) and then
density increases with calcination temperature. Such observation is common among
flash calcined kaolinites [Bridson et al., 1985; Slade and Davies, 1991], even soak cal-
cined kaolinites display similar trend in density [Grim, 1953]. When kaolinite is exposed
to temperatures 900 – 1200 K, it tends to lose the structural hydroxyls to form a poorly
ordered structure, metakaolinite. This material is characterized by amorphicity and low
density [Davies, 1986]. As calcination temperature increases, a slow increase in density
is observed which is due to the rearrangement of alumina and silica structures to form a
Chapter 5. Result and Discussion 52

Figure 5.2: Comparison between GSC tests and model predictions: density of the
clay samples

dense mullite phase. The X-ray pattern observed for the calcined products in Figure 4.11
is a good indication for the formation of mullite phase at higher calcination temperature
(MK1273WB ).

The degree of dehydroxylation, α which implies the amount of crystal water removed
after the kaoilinite rich clay material has been exposed to various calcination condi-
tions, is also another parameter that is used to compare experimental data with model
prediction. The degree of dehydroxylation of the calcined clay at different calcination
temperatures is compared with model results. Figure 5.3 presents a comparison between
the experiments and model prediction.

Figure 5.3: Comparison between GSC tests and model predictions: Degree of dehy-
droxylation
Chapter 5. Result and Discussion 53

Despite the presence of some uncertainties associated with the operation of GSC such
as temperature drop along the GSC, the above comparison is generally found to be in
good agreement.

Other experimental data that need to be discussed is, the composition of phases after the
feed clay material has been exposed to different calcination conditions. As it is described
in the previous chapter, XRD and XRF have been a prominent tools to estimate the
mineralogical composition and the amount of phases in the clay sample. The estimation
is made by Rietveld refinement of randomly oriented samples using the known miner-
alogy of the feed clay material as input. As shown in Figure 4.11, it is clear that the
diffraction signals from kaolinite decreases as the the calcination temperature increase.
This is due to transformation of kaolinite into other phases, mainly to metakaolinite.
Unfortunately metakaolinite is amorphous phase and is not shown in the XRD pattern
but can be estimated indirectly. Another interesting observation from the XRD pat-
tern is, the formation of mullite is not shown for all calcined products, except for those
samples exposed to temperatures about 1273 K in the presence of secondary burners
(MK1273WB ). This implies that all other calcination temperatures are safe but may not
cause complete transformation of kaolinite to metakaolinite at the given residence time.

Table 5.2 compares the composition of phases in the calcined clay materials with model
prediction. A detailed analysis on the comparison of phase composition between the
experiments and model can be found in Paper 2. Such comparison must account the un-
certainties associated with XRD method, such as, use of anatase as a reference, weighing
of samples, peak-overlap, low signal/noise ratios and poor simulations in Rietveld XRD.

Table 5.2: Experimentally estimated and model-predicted composition of clay samples

Kaolinite (%) Quartz (%) Amorphous (%) Mullite (%)


Sample
Experiment Model Experiment Model Experiment Model Experiment Model
Feed 90-95 94 5-10 6 - - 0 0
MK964NB 10* 31.1 6 6.6 84* 63 0 0
MK1058NB 1 2.5 8 7 91 90.4 0 0
MK1126NB 0 1.23 9 7.7 91 90 0 0
MK1214NB 0 0 9 8.9 91 87 0 1.05
MK1073WB 0 1.2 8 7 92 90.1 0 4.02
MK1273WB 0 0 10 13.7 74 71.76 16 14.4
∗Thermal analysis suggests 27% of kaolinite for MK964NB which leaves the amorphous content to be 66%

A cross-comparison between the C++ model and gPROMS results is also made, as a
kind of model validation effort. The agreement between the two modeling tools verifies
the mathematical approach of the model. Figure 5.4 illustrates the comparison of model
output in C++ and gPROMS, where the temperature profile and the conversion are
compared. Detailed results based on gPROMS ModelBuilder can be found in Paper 3.
Chapter 5. Result and Discussion 54

Figure 5.4: The comparison between C++ and gPROMS results (A) temperature
profile at particle center/surface (B) model predicted conversion profile

5.3 Detailed model prediction

5.3.1 Baseline case results

The following detailed explanation on model results is based on the calcination conditions
given in Table 5.3. Since most physical properties are adopted from the bulk properties
of kaolinite clay powder, a spherical kaolinite clay particle having a diameter of 100 µm
is assumed. The model results explained here are similar to those explained in Paper 1
and Paper 2, except that larger particle size and “average” kinetic parameters are used
to simulate the current results.
Table 5.3: The initial properties of kaolinite clay and calcination conditions

Variables Values
Initial density of feed clay particle 2654 kg/m3
mass fraction of kaolinite in particle 93%
mass fraction of silica in particle 6.1%
mass fraction of moisture in particle 0.9%
Initial temperature of clay particle, T0 298 K
Clay particle diameter, Dp 100 µm
Biot number, Bi hT Dp /6k
Pore diameter, dp 3 × 10−7 m
Porosity, ε 0.5
Ambient gas composition Air (77 wt.% N2 and 23wt.% O2 )
Ambient gas temperature, T∞ 1273 K
Radiation temperature, Trad Trad = T∞
Ambient gas pressure, Patm 101,325 Pa
Chapter 5. Result and Discussion 55

When a kaolinite particle is plunged into a hot air at temperature 1273 K, it undergoes
a sequence of transformations where the heat and mass transfer profile is given in Fig-
ure 5.5. As shown in the figure, the particle surface temperature is remarkably higher
than the temperature in the center. This is due to convective and radiative transfer of
heat from the surrounding to the particle surface which then transfers to the center by
conduction until the particle attains isothermal condition.

T (Surface) T (Center)
Y_H2O (Center) Y_H2O (Surface)

1400 1

1200

Mass fraction, Y_H2O [‐]
0.8
Temperature, T[K]

1000
0.6
800
0.4
600

400 0.2

200 0
0 0.1 0.2 0.3 0.4 0.5
Time, t[s]

Figure 5.5: Model predicted profiles in temperature and water vapor at particle
center/surface

As the clay particles experiences temperature gradient within the particle, it also dis-
plays a distinctly different mass fraction of water vapor across the particle radius. The
two peaks observed on the above figure correspond to evaporation of free water and
dehydroxylation, respectively. Since evaporation and dehydroxylation reactions are en-
dothermic reactions, the energy demand by these reactions cause the temperature profile
to be damped; at this instance, the the temperature and vapor mass fraction showed
a clear gradient between the surface and center of the particle. Moreover, the intense
release of water vapor during dehydroxylation gives rise to outward flow of vapor in-
side the solid matrix that offsets the inward heat conduction giving rise to maximum
temperature difference between the particle surface and center during the two reaction
periods.

Figure 5.6 illustrates the variation of kaolinite density as it transforms to metakaolinite


and mullite. The density of kaolinite in the clay particle decreases as the dehydroxylation
reaction starts. Kaolinite is completely depleted at the full dehydroxylation about 0.22
seconds, and transformed to metakaolinite. Further holding on the clay particle in
Chapter 5. Result and Discussion 56

the hot air will lead to the transformation of metakaolinite into unwanted products,
e.g., mullite. The particle conversion is deduced based on changes in particle mass.
Accordingly, the conversion, X is computed as,

(mkl + mf w )
X =1− (5.1)
(m0kl + m0f w )

where, mkl and mf w represent the mass of kaolinite and free water in the clay particle at
time t, while m0kl and m0f w denote their initial values in the raw clay particle, respectively.

As shown in Figure 5.6, the particle conversion is plotted against time. The clay particle
attains complete conversion in about 0.22 seconds, at this time the clay product contains
optimum amount of metakaolinite. The 14% of mass loss belongs to the amount of crystal
water liberated during dehydroxylation.

Figure 5.6: Model predicted density of solid species and conversion of the particle

The fact that koalinite undergoes a variation in density during flash calcination has been
experimentally demonstrated [Davies, 1986; Slade and Davies, 1991]. The time at which
the calcined product exhibit lower density has been assigned to the optimum amount
of metakaolinite. According to Bridson et al. [1985], during calcination, kaolinite clay
undergoes a change in density in a trend that depends on the degree of dehydroxylation.
For such cases, an empirical trend is proposed as;

ρ = ρkl (1 − βα) (5.2)


Chapter 5. Result and Discussion 57

where, ρ is the density of flash calcined product, ρkl is the density of the kaolinite
sample, β is the mass ratio of water in the kaolinite sample (≈ 14%) and α is the degree
of dehydroxylation. The implication of this correlation is, for kaolinite clay having initial
density 2630 kg/m3 , the calcined clay material could have a density as low as 2200 kg/m3 ;
provided that the volume of the clay material is assumed constant. In our model, using
similar kaolinite properties such as, the density of kaolinite and β, as stated in Eq.
5.2, the same trend was observed in density as the empirical correlation established by
Bridson et al. [1985]. In reality, flash calcines undergo a little swelling in volume and
the density of the calcines is observed lower than the empirical trend. Flash calcination
experiments carried out by Slade et al. [1992] are chosen for comparison purposes, where
kaolinite clay sample is flash calcined at 1273 K for 0.5 seconds in nitrogen atmosphere.
Different calcines are produced at different degrees of dehydroxylation by controlling the
residence time. The calcines exhibit lower mean density than the suggested empirical
trend; this is due to the presence of voids in their structure. Figure 5.7 illustrates
the variation in density of kaolinite with degree of dehydroxylation for flash calcination
experiments performed by Slade et al. [1992], the trend suggested by Bridson et al. [1985]
and the model prediction. The agreement between model and the empirical trend is due
to the assumption of negligible volume change during calcination.

2600
Density, ρ [kg/m3]

2400

2200
Experimental (Slade et al., 1992)
Emperical (Bridson et al., 1985)
2000
Model (this work)

1800
0 0.2 0.4 0.6 0.8 1
Degree of dehydroxylation, α

Figure 5.7: The changes in density of kaolinite as a function of degree of dehydrox-


ylation during calcination: comparison between experiment, empirical correlation and
model prediction.

In an attempt to locate the optimum amount of metakaolinite, Figure 5.8 illustrates the
variation in density of the kaolinite model against time. The density of the calcined clay
Chapter 5. Result and Discussion 58

products, ρ is calculated based on the mass fraction of individual solid species and their
respective true density, which is given as,
 X Y −1
i
ρ= (5.3)
ρi
i

where, Yi and ρi represent the mass fraction and material density of ith solid component
in the clay particle. The material densities used in this model are 2589, 2550, 3100, 2650,
2200 and 2270 kg/m3 for kaolinite, metakaolinite, mullite, quartz, silica and cristobalite,
respectively [Al-Akhras, 2006; Database, 2013].

Figure 5.8: The variation in model predicted density of the calcined product as a
function of time

It is evident from the graph above that for the calcination conditions specified above,
most of the kaolinite particle gets dehydroxylated between 0.2–0.3 seconds. This obser-
vation is consistent with literature [Cadoret, 2005; Salvador and Davies, 1994].

The significance of convection, diffusion and chemical reaction during calcination has
been examined for the particle model by using Peclet-Thiele mapping [Cardoso and
Rodrigues, 2007; Lopes et al., 2009]. In order to assess the relative effect of chemical
kinetics and mass transport through convection and diffusion, dimensionless parameters
are used. Thiele modulus, φ is one of these parameters used for physical interpretation of
the two time scales involved in the system; reaction and diffusion. Its value is computed
as, s
Rp Sg
φ= (5.4)
3 ερg Deff

The influence of convection on the mass transfer of vapor can also be evaluated based
on the Peclet mass transfer number, P em which compares the ratio of convective to
Chapter 5. Result and Discussion 59

diffusive mass transfer as:


Rp U
P em = (5.5)
3Deff
The regime diagram suggested by Cardoso et al. [Cardoso and Rodrigues, 2007] and
Lopes et al. [Lopes et al., 2013, 2009] is based on the values of Peclet and Thiele modules.
Accordingly, the different regimes illustrated in Figure 5.9 characterize the interaction
between chemical reaction, diffusion and intra-particle convection during the process.
Thus, when the dominant phenomena are convection and diffusion, the flow is charac-
terize by small values of Thiele modulus (φ2 /P em  1). For smaller Peclet numbers
(P em  1), and (φ2 ≈ 1), diffusion and reaction are the dominant mechanisms. Inter-
mediate values of the Thiele modulus (φ2 /P em ≈ 1) illustrate convection and reaction
are limiting during the process. As shown in the figure (dashed lines), the P em values
lie between 0 and 0.8 while the φ2 values lie between 0 and 1.1. Thus, the flow regime of
kaolinite dehydroxylation falls in the region where Diffusion-Convection-Reaction plays
significant role. This ensures the importance of incorporating the diffusion, convection
and reaction terms in the model. Based on the regime indicated for calcination process
(dashed line) in the Figure 5.9, it can be generally suggested that diffusion is a dominant
phenomena during calcination kaolinite rich clays. To be more specific, the small values
of Thiele modulus (0 ≤ φ2 <1.1) and Peclet number (0 ≤ P em <0.8) indicate convec-
tion, diffusion and reaction kinetics have a major impact during the process of thermal
calcination, with diffusion being dominant over others.

Figure 5.9: Regime diagram for convection, diffusion and chemical kinetics repro-
duced from Literature [Lopes et al., 2009] showing a specific region that belong to the
calcination process of kaolinite clay (dashed line)
Chapter 5. Result and Discussion 60

5.3.2 Sensitivity analysis

The parameter values and assumptions of any model are subject to change. The main
aim of performing sensitivity analysis is to investigate the potential change of parameters
and their impact on the conclusion to be drawn from the model. In this study, a
sensitivity analysis is performed to examine the effect of calcination conditions (e.g.,
calcination temperature, time), clay particle properties (e.g., particle size) and some
uncertain model inputs (e.g., kinetic parameters) on the final property of the calcined
clay material. Detailed sensitivity analysis can be found in Paper 3.

Calcination temperature and time

As it is mentioned earlier, the thermal transformation of kaolinite into different phases is


entirely dependent on calcination temperature. As the calcination temperature increase,
there is high risk of formation of the undesired product called mullite that influence
the pozzolanic property of the final product negatively. In order to investigate the
sensitivity of calcination temperature on the property of the final product, the gas
temperature has been varied from 1073 K to 1373 K and the impact on conversion profile
is studied. Figure 5.10(A) depicts the effect of calcination temperature on the conversion
of the kaolinite clay model. It is shown that, the higher the temperature, the faster the
conversion to metakaolinite. However, the amount of metakaolinite is observed to be
depleted as temperature and residence time increase. This is due to the transformation of
metakaolinite into undesired phase. Figure 5.10(B) illustrates the impact of calcination
temperature and residence time on the amount of metakaolinite. Thus, it is apparent
that the right residence time and calcination temperature are extremely important to
obtain maximum concentration of metakaolinite in the final calcined product.

Figure 5.10: Sensitivity of calcination temperature and time on (A) model predicted
conversion (B) model predicted mass fraction of metakaolinite
Chapter 5. Result and Discussion 61

Particle size

Clay is composed of a number of tiny particles with several dimensions as shown in the
experimental section. with this regard, sensitivity analysis on the particle size gives an
overall idea on how the different particles behave during calcination. To investigate the
effect of particle size on model results, different particle sizes that range from 25 µm to
150 µm are assumed and their impact on the conversion is studied. Figure 5.11 illustrates
the impact of particle size on conversion.

25µm 50µm 100µm 150µm

1.0

0.8
Conversion

0.6

0.4

0.2

0.0
0 0.1 0.2 0.3 0.4 0.5
Time, t[s]

Figure 5.11: Sensitivity of particle size on model predicted conversion

As shown in the figure above, the conversion profile is a strong function of particle
size. Thus, smaller particles are observed to attain full conversion rapidly, whereas
larger particles of diameter 150 µm, the intra-particle heat and mass transfer resistances
become more important than the external heat and mass transfer resistance and this
delays the time of full conversion.

Kinetic parameters

Since the expression of reaction rates is based on Arrhenius type of dependence which
is interpreted in terms of kinetic parameters (E and A), the influence of these param-
eters on thermal calcination of the kaolinite model is unquestionable. A number of
kinetic parameters have been reported in the literature, specially for dehydroxylation
reaction step. Table 5.4 summarizes some selected values of kinetic parameters given for
dehydroxylation of kaolinite from the literature.
Chapter 5. Result and Discussion 62

Table 5.4: A summary of kinetic parameters for kaolinite dehydroxylation from liter-
ature

E(kJ/mol) A(s− 1) Method Source Reference


177 4.57 × 108 Thermal Analysis, DTA Florida, USA [Kissinger, 1956]
193 1.0 × 109 Thermal Analysis, TGA Georgia, USA [Bellotto et al., 1995]
196 9.6 × 108 Thermal Analysis, TGA India [Saikia et al., 2002]
195 8.58 × 1014 Thermal Analysis, DTG Rep. Czech [Ptáček et al., 2011]
163 2.0 × 1012 Thermal Analysis, TGA USA [Levy and Hurst, 1993]
180 6.3 × 109 Thermal Analysis, TGA EU this work

Here, the influence of the above mentioned kinetic parameters on the degree of con-
version of kaolinite model has been investigated and compared with the experimentally
determined kinetic parameters in this study. Figure 5.12 illustrates the impact of ki-
netic parameters on the conversion profile of the model. Consequently, the sensitivity of
kinetic parameters can be judged from the different conversion profiles observed in the
figure.

Ptacek et.al. Levy et.al. Experiment


Kissinger Belloto et.al. Saikia et.al.
1

0.8
Conversion

0.6

0.4

0.2

0
0 0.1 0.2 0.3 0.4 0.5
Time, t [s]

Figure 5.12: The sensitivity of kinetic parameters based on literature and their impact
on the conversion profile of the model
Chapter 6

Conclusion

This section brings the summary and conclusions made out of the study. Recommenda-
tions are also made to further improve the existing research work based on more reliable
inputs and suggestions on the particle model.

6.1 Final remarks

Thermal calcination of kaolinite has been of current interest in cement and concrete
industry. Based on its application for the intended use, the calcined material needs to
have high concentration of the pozzolanic active material, metakaolinite. Besides, flash
calcined metakaolinite displays typical properties that enhance the materials pozzolanic
property. As flash calcination is a very quick process that could yield a calcined product
in a fraction of seconds, modeling of such processes enable us to capture and understand
the inter-particle processes during transformation.

In this work, thermal calcination kaolinite rich clay particles is studied by using both
modeling and experimental approaches. Both modeling and experimental results show
a good agreement.

A comprehensive mathematical model has been successfully developed for clay particle
calcination, which sufficiently addresses the key intra-particle processes and the particle-
ambient flow interaction. The model reliably predicts favorable production path that
may able to achieve an optimum amount of the desired product, namely metakaolin-
ite. Based on the kinetic, physical and thermodynamic data used in the model, when
kaolinite rich clay particles are plunged into a hot gas atmosphere, the clay particles are
observed to dehydroxylate quickly. The speed at which the kaolinite clay is converted
into metakaolinite depends on the calcination temperature, particle size and other inputs

63
Chapter 6. Conclusion 64

such as kinetic parameters. Simulation of kaolinite particle model in different calcination


temperature has predicted interesting results that may give a full picture on the opera-
tion conditions of a flash calciner. Generally, there are three entirely correlated factors
that may influence the process: calcination temperature, particle size and the yield of
metakaolinite. In order to achieve optimum amount of metakaolinite, a compromise
is important among the above parameters. Clay particles exposed to high calcination
temperatures (>1273 K) may exhibit not only faster conversion but also faster phase
transformation into undesired material. Again, lower temperatures (<1073 K) may re-
quire longer residence time and may not have the risk of phase transformation. Based
on these observations, calcination temperatures 1173 K–1273 K may hold a reasonable
tradeoff in the amount of metakaolinite and residence time. At such calcination tempera-
tures, even though the model predicts lower residence time (0.2 – 0.3 s), when calcination
continues for half a second (0.5 s) the calcined material still contains sufficient amount
of metakaolinite enough to be used as an input in concrete production.

From the modeling perspective, the physical and thermodynamic properties that char-
acterize clay particles are found to limit simulation results. The complex nature of clays
is also another factor that may result in diverse values of kinetic parameters. The sensi-
tivity analysis on kinetic parameters revealed the influence on dehydroxylation process.
In such cases of uncertainty, tuning the kinetic parameters based on experimental results
can provide a good estimation of their values.

Flash calcination experiments in a gas suspension calciner are presented for a residence
time of 0.5 seconds. Under different calcination temperatures, the composition of the
calcined material is experimentally examined. In all cases, the calcined product is ob-
served to comprise high amount of the required material, metakaolinite. Yet, optimum
amount of metakaolinite is noticed at higher temperatures; except for temperatures at
1273 K in the presence of secondary burners. This observation is in agreement with
model prediction. However, the need for a stable temperature inside the calciner is
important to mention.

6.2 Future work

As there is only a little work reported in the literature about modeling calcination of
kaolinite clay, some recommendations that might help to further improve the results of
this study are described below.

In order to gain more fundamental insight on the calcination process, the reaction rate on
the more influencing reaction step, which is dehydroxylation, might need more refinement
Chapter 6. Conclusion 65

and better expression. Owing to the sensitivity of the calcination process towards kinetic
parameters and the difficulty of determining these variables, as briefly stated in this work,
rate expressions that are based on Arrhenius equation could contribute to the major
share for such uncertainties. A new paradigm shift may be necessary in expressing the
reaction rates for solid-gas reactions as described in [Favergeon et al., 2013; Michèle
et al., 2011]. These expressions may somehow solve the doubts on predicted properties
that vary significantly with kinetic parameters.

From the perspective of the particle model, the first question that might appear could
be whether one dimensional approach is sufficient to model the complex structure and
nature of clays. Two dimensional or fully three dimensional structure of kaolinite clay
particles could be more appropriate to address the intra-particle transport phenomenon
and any structural change, if proper description of particle shape is available.

In the entire SCM project, a lot of good efforts are made on the processes at the down-
stream side of the calciner, e.g., comprehensive characterization of the calcines obtained
under various calcination condition, and concrete tests by using the calcined clay in con-
crete. In this thesis, models for single clay particle calcination are successfully developed
and validated. However, all these are far from being enough, in terms of cost-effective
innovative design of smart, energy-efficient large-scale gas suspension clay calciners. To
achieve this, advanced CFD simulation of the reacting particulate flow system in the
calciner must be performed, in which all the highly coupled sub-processes (e.g., com-
plicated particulate flow, heat transfer, clay particle conversion, particle agglomeration
and so on) must be appropriately addressed. The CFD simulation of the entire calciner
must be somehow validated (e.g., by experimental data) to assure that a reliable CFD
modeling capability for such a gas suspension flash calcination process is established.
Only after that, a true understanding of the details in the calciner and the impacts of
various factors can be reliably achieved (e.g., via CFD model-based sensitivity study),
and smart, energy-efficient large-scale gas suspension calciners can be designed and op-
timized in a reliable and cost-effective way (i.e., mainly based on the CFD capability,
with proper combination of experience and a few tests and adjustment).
Bibliography

Abramzon, B. and W. Sirignano (1989). Droplet vaporization model for spray combus-
tion calculations. Int. J. Heat Mass Transf. 32 (9), 1605–1618.

Al-Akhras, N. (2006). Durability of metakaolin concrete to sulfate attack. Cem. Concr.


Res. 36, 1727–1734.

Allison, E. (1955). Quantitative thermal analysis of clay minerals. Clay Miner. Bull 2,
242–254.

Ambroise, J., M. Murat, and J. Péra (1985). Hydration reaction and hardening of cal-
cined clays and related minerals. V. Extension of the research and general conclusions.
Cem. Concr. Res. 15, 261–268.

Arikan, M., K. Sobolev, T. Ertün, A. Yeğinobali, and P. Turker (2009). Properties of


blended cements with thermally activated kaolin. Constr. Build. Mater. 23 (1), 62–70.

Augis, J. and J. Bennett (1978). Calculation of the Avrami parameters for heterogeneous
solid state reactions using a modification of the Kissinger method. J. Therm. Anal.
Calorim. 13, 283–292.

Baumann, W., A. Leineweber, and E. J. Mittemeijer (2010). Failure of Kissinger (-like)


methods for determination of the activation energy of phase transformations in the
vicinity of the equilibrium phase-transformation temperature. J. Mater. Sci. 45 (22),
6075–6082.

Bellotto, M., A. Gualtieri, G. Artioli, and S. Clark (1995). Kinetic study of the kaolinite-
mullite reaction sequence. Part I: kaolinite dehydroxylation. Phys. Chem. Miner. 22,
207–214.

Benitez, J. (2009). Principles and modern applications of mass transfer operations (Sec-
ond ed.). New jersy: John willey & sons.

Bergman, T., A. Lavine, F. Incropera, and D. DeWitt (2011). Fundamentals of heat and
mass transfer (Seventh ed.). Jefferson city, USA: Jonh Wiley & Sons.

67
Bibliography 68

Bhatia, S. (1985). On the pseudo steady state hypothesis for fluid solid reactions. Chem.
Eng. Sci. 40 (5), 868–882.

Bish, D. and J. Post (1993). Quantitative mineralogical analysis using the Rietveld
full-pattern fitting method. Am. Mineral. 78, 932–940.

Bolz, R. and G. Tuve (1976). Handbook of Tables for Applied Engineering Science
(Second ed.). New York: CRC press.

Borgwardt, R. H. (1985). Calcination kinetics and surface area of dispersed limestone


particles. AIChE J. 31 (1), 103–111.

Boswell, P. (1980). On the calculation of activation energies using a modified Kissinger


method. J. Therm. Anal. Calorim. 18, 353–358.

Bridson, D., T. Davies, and D. Harrison (1985). Properties of flash-calcined kaolinite.


Clays Clay Miner. 33 (3), 258–260.

Brindley, G. and M. Nakahira (1958). A new concept of the transformation sequence of


kaolinite to mullite. Nature (4619), 1333–1334.

Brindley, G., J. Sharp, J. Petterson, and B. Narahari (1967). Mechanism of dehydrox-


ylation processes, I. Temperature and vapor pressure dependence of dehydroxylation
of kaolinite. Am. Mineral. 52 (65), 201–211.

Bryden, K. M. and M. J. Hagge (2003). Modeling the combined impact of moisture and
char shrinkage on the pyrolysis of a biomass particle. Fuel 82 (13), 1633–1644.

Cabrera, J. and M. Eddleston (1983). Kinetics of dehydroxylation and evaluation of the


crystallinity of kaolinite. Thermochim. Acta 70, 237–247.

Cadoret, G. (2005). Dehydroxylated aluminium silicate based material, process and


installation for the manufacture thereof, US patent 2005/0039637 A1, Feb. 24, 2005.

Cardoso, S. S. and A. E. Rodrigues (2007). Convection, diffusion and reaction in a


nonisothermal, porous catalyst slab. AIChE journal 53 (5), 1325–1336.

Cassagnabère, F., M. Mouret, G. Escadeillas, P. Broilliard, and A. Bertrand (2010).


Metakaolin, a solution for the precast industry to limit the clinker content in concrete:
Mechanical aspects. Constr. Build. Mater. 24 (7), 1109–1118.

Chen, D., X. Gao, and D. Dollimore (1993). A generalized form of the Kissinger equation.
Thermochim. Acta 215, 109–117.

Chen, Y., M. Wang, and M. Hon (2003). Transformation kinetics for mullite in kaolin-
Al2O3 ceramics. J. Mater. Res 18 (6), 1355–1362.
Bibliography 69

Chung, F. H. and D. K. Smith (1999). Industrial Applications of X-Ray Diffraction.


NY, USA: CRC Press.

Criado, J., A. Ortega, C. Real, and E. Torres De Torres (1984). Re-Examination of the
Kinetics of the Thermal Dehydroxylation of Kaolinite. Clay Miner. 19 (4), 653–661.

Dam, T. (2013). Supplementary Cementitious Materials and Blended Cements to Im-


prove Sustainability of Concrete Pavements. Natl. Concr. Pavement Technol. Center.
Iowa state Univ., http://www.cptechcenter.org/technical–library/docu, online: ac-
cessed on March 2014.

Database (2013). Mineralogic database-Mineral Collecting, Localities, Mineralogy


database-Mineral Photos and Data. http://www.mindat.org/, Accessed on January,
2013.

Davies, T. (1986). Flash dehydroxylation of kaolinite: the effect of heating rate on the
properties of the calcine. J. Mater. Sci. Lett. 5, 186–187.

Diamond, S. (1970). Pore Size Distributions in Clays. Clays Clay Miner. 18 (1), 7–23.

Dion, P., J. Alcover, F. Bergaya, A. Ortega, P. L. Llewellyn, and F. Rouquerol (1998).


Kinetic study by controlled-transformation rate thermal analysis of the dehydroxyla-
tion of kaolinite. Clay Miner. 33 (2), 269–276.

Favergeon, L., J. Morandini, M. Pijolat, and M. Soustelle (2013). A General Approach


for Kinetic Modeling of Solid-Gas Reactions at Reactor Scale: Application to Kaolinite
Dehydroxylation. Oil Gas Sci. Technol. – Rev. IFP Energies Nouv. 68 (6), 1039–1048.

Fernandez, R., F. Martirena, and K. L. Scrivener (2011). The origin of the pozzolanic
activity of calcined clay minerals: A comparison between kaolinite, illite and mont-
morillonite. Cem. Concr. Res. 41 (1), 113–122.

Fidaros, D., C. Baxevanou, C. Dritselis, and N. Vlachos (2007). Numerical modelling


of flow and transport processes in a calciner for cement production. Powder Tech-
nol. 171 (2), 81–95.

Georgakis, C., C. Chang, and J. Szekely (1979). A changing grain size model for
gas—solid reactions. Chem. Eng. Sci. 2 (4), 1072–1075.

Gerardin, C. and S. Sundaresan (1994). Structural investigation and energetics of mullite


formation from sol-gel precursors. Chem. Mater. 6 (2), 160–170.

Grim, R. E. (1953). Clay mineralogy. New York: McGRAW-HILL.


Bibliography 70

Hankock, J. D. and J. H. Sharp (1972). Method of Comparing Solid-State Kinetic Data


and Its Application to the Decomposition of Kaolinite, Brucite, and BaCO3. J. Am.
Ceram. Soc. 55 (2), 74–77.

He, C., E. Makovicky, and B. Osbæ ck (1994). Thermal stability and pozzolanic activity
of calcined kaolin. Appl. Clay Sci. 9, 165–187.

Hewlett, P. (1998). Lea’ s Chemistry of Cement and Concrete (4th ed.). Burlington,
USA: Elsevier Science and Technology Books.

Holm, J. (2001). Kaolinites–mullite transformation in different Al2O3–SiO2 systems:


Thermo-analytical studies. Phys. Chem. Chem. Phys., 10–13.

Holm, J., O. Kleppa, and E. Westrum (1967). Thermodynamics of polymorphic trans-


formations in silica. Thermal properties from 5 to 1070 K and pressure-temperature
stability fields for coesite and stishovite. Geochim. Cosmochim. Acta 31, 2289–2307.

Horvath, I. (1985). Kinetics and Compensation effect in kaolinite Dehydroxylation.


Thermochem. Acta 85, 193–198.

Hu, N. and A. W. Scaroni (1996). Calcination of pulverized limestone particles under


furnace injection conditions. Fuel 75 (2), 177–186.

Humphreys, K. and M. Mahasenan (2002). Towards a sustainable cement industry sub-


study 8: climate change. http://www.wbcsdcement.org/pdf/battelle/final report8.pdf ,
Online; accessed march 2014.

IEA (2009). Cement Technology Roadmap 2009. International Energy Agency


(IEA)/World Business Council for Sustainable Development(WBCSD) report. Tech-
nical report.

Khatib, J., E. Negim, and E. Gjonbalaj (2012). High volume metakaolin as cement
replacement in mortar. World J. Chem. 7 (1), 7–10.

Kissinger, H. (1957). Reaction kinetics in differential thermal analysis. Anal. Chem. 29,
1702–1706.

Kissinger, H. E. (1956). Variation of peak temperature with heating rate in differential


thermal analysis. J. Res. Natl. Bur. Stand. (1934). 57 (4), 217–21.

Knovel (2013). Technical engineering reference information, http://why.knovel.com/,


accessed on july, 2013.

Lahiri, A. (1980). The effect of particle size distribution on TG. Thermochim. Acta 40,
289–295.
Bibliography 71

Levy, J. and H. Hurst (1993). Kinetics of dehydroxylation, in nitrogen and water vapour,
of kaolinite and smectite from Australian Tertiary oil shales. Fuel 72 (6), 873–877.

Li, C., H. Sun, and L. Li (2010). A review: The comparison between alkali-activated
slag (Si+Ca) and metakaolin (Si+Al) cements. Cem. Concr. Res. 40 (9), 1341–1349.

Llópiz, J., M. Romero, A. Jerez, and Y. Laureiro (1995). Generalization of the Kissinger
equation for several kinetic models. Thermochim. Acta 256, 205–211.

Lopes, J., M. Alves, M. Oliveira, S. Cardoso, and A. Rodrigues (2013). Internal mass
transfer enhancement in flow-through catalytic membranes. Chem. Eng. Sci. 104,
1090 – 1106.

Lopes, J., S. Cardoso, and A. Rodrigues (2009). Convection, diffusion, and exothermic
zero order reaction in a porous catalyst slab: Scaling and perturbation analysis. AIChE
J. 55 (10).

Lothenbach, B., K. Scrivener, and R. Hooton (2011). Supplementary cementitious ma-


terials. Cem. Concr. Res. 41 (12), 1244–1256.

Lu, H., W. Robert, G. Peirce, B. Ripa, and L. L. Baxter (2008). Comprehensive Study
of Biomass Particle Combustion. Energy & Fuels 22 (4), 2826–2839.

Massazza, F. (1993). Pozzolanic cements. Cem. Concr. Compos. 15 (1993), 185–214.

Mehta, B. Y. P. K. (2002). Greening of the Concrete Industry for Sustainable Develop-


ment. Concr. Int., 23–28.

Meinhold, R., H. Atakul, and T. Davies (1992). Flash calcines of kaolinite: kinetics
of isothermal dehydroxylation of partially dehydroxylated flash calcines and of flash
calcination itself. J. Mater. Chem. 38 (9), 913–921.

Meinhold, R. and S. Salvador (1994). A comparison of the kinetics of flash calcination


of kaolinite in different calciners. IChemE 72 (part A), 105–113.

Michèle, P., F. Loı̈c, and S. Michel (2011). From the drawbacks of the Arrhenius-f(α) rate
equation towards a more general formalism and new models for the kinetic analysis
of solid–gas reactions. Thermochim. Acta 525 (1-2), 93–102.

Michot, A., D. S. Smith, S. Degot, and C. Gault (2008). Thermal conductivity and spe-
cific heat of kaolinite: Evolution with thermal treatment. J. Eur. Ceram. Soc. 28 (14),
2639–2644.

Mikulčić, H., E. von Berg, M. Vujanović, P. Priesching, L. Perković, R. Tatschl, and


N. Duić (2012). Numerical modelling of calcination reaction mechanism for cement
production. Chem. Eng. Sci. 69 (1), 607–615.
Bibliography 72

Murray, H. H. (1963). Industrial applications of kaolin. In Tenth Natl. Conf. clays clay
Miner., pp. 291–298.

Murray, P. and J. White (1955). Kinetics of clay dehydration. Clay Min. 2, 255–264.

Oh, M. and C. Pantelides (1996). A modelling and simulations language for combined
lumped and distributed parameter systems. Comput. chem.Engng 20, 661–633.

Opfermann, J. and H. Flammersheim (2003). Some comments to the paper of JD Sewry


and ME Brown:“Model-free” kinetic analysis? Thermochim. Acta 397, 1–3.

Ortega, A., M. Macias, and F. J. Gotor (2010). The Multistep Nature of the Kaolinite
Dehydroxylation: Kinetics and Mechanism. J. Am. Ceram. Soc. 93 (1), 197–203.

Pantelides, C. and P. Barton (1993). Equation-oriented dynamic simulation current


status and future perspectives. Comput. Chem. Eng. 17, S263–S285.

Patankar, S. v. (1980). Numerical heat transfer and fluid flow. USA: Hemispher Pub-
lishing Corporation.

Patisson, F., M. G. François, and D. Ablitzer (1998). A non-isothermal, non-equimolar


transient kinetic model for gas-solid reactions. Chem. Eng. Sci. 53 (4), 697–708.

Pijolat, M., F. Valdivieso, and M. Soustelle (2005). Experimental test to validate the
rate equation “dα/dt=kf(α)” used in the kinetic analysis of solid state reactions.
Thermochim. Acta 439 (1-2), 86–93.

Prasad, M., K. Reid, and H. Murray (1991). Kaolin: processing, properties and appli-
cations. Appl. Clay Sci. 6 (2), 87–119.

PSE (2013). gPROMS Model Developer’s Guide, www.psenterprise.com.

Ptáček, P., M. Křečková, F. Šoukal, T. Opravil, J. Havlica, and J. Brandštetr (2012).


The kinetics and mechanism of kaolin powder sintering I. The dilatometric CRH study
of sinter-crystallization of mullite and cristobalite. Powder Technol. 232, 24–30.

Ptáček, P., F. Šoukal, T. Opravil, J. Havlica, and J. Brandštetr (2011). The kinetic
analysis of the thermal decomposition of kaolinite by DTG technique. Powder Tech-
nol. 208 (1), 20–25.

Ptáček, P., F. Šoukal, T. Opravil, M. Nosková, J. Havlica, and J. Brandštetr (2010). The
non-isothermal kinetics analysis of the thermal decomposition of kaolinite by Effluent
Gas Analysis technique. Powder Technol. 203 (2), 272–276.

Ranz, W. and W. Marshall (1952). Evaporation from drops. Chem. Eng. Prog. 48,
141–146.
Bibliography 73

Redfern S. (1987). kinetics of kaoline dehydroxyaltion of kaolinite. Clay Miner. 22,


447–456.

Rehan, R. and M. Nehdi (2005). Carbon dioxide emissions and climate change: policy
implications for the cement industry. Environ. Sci. Policy 8 (2), 105–114.

Reinecke, S. and B. E. Sleep (2002). Knudsen diffusion, gas permeability, and water
content in an unconsolidated porous medium. Water Resour. Res. 38 (12), 1–15.

Robie, R. A., B. S. Hemingway, and J. R. Fisher (1979). Thermodynamic Properties


of Minerals and Related Substances at 298.15 K and 1bar pressure and at higher
temperatures. Washington D.C: United states government printing office.

Sabir, B., S. Wild, and J. Bai (2001). Metakaolin and calcined clays as pozzolans for
concrete: a review. Cem. Concr. Compos. 23 (6), 441–454.

Said-Mansour, M., E.-H. Kadri, S. Kenai, M. Ghrici, and R. Bennaceur (2011). Influence
of calcined kaolin on mortar properties. Constr. Build. Mater. 25 (5), 2275–2282.

Saikia, N., P. Sengupta, P. K. Gogoi, and P. C. Borthakur (2002). Kinetics of dehydrox-


ylation of kaolin in presence of oil field effluent treatment plant sludge. Appl. Clay
Sci. 22 (3), 93–102.

Salvador, S. (1995). Pozzolanic properties of flash-calcined kaolinite: a comparative


study with soak-calcined products. Cem. Concr. Res. 25 (1), 102–112.

Salvador, S. and T. Davies (1994). Modeling of combined heating and dehydroxylation


of kaolinite particles during flash calcination; production of metakaolinite. Process.
Adv. Mater. 9, 128–135.

San Nicolas, R., M. Cyr, and G. Escadeillas (2013). Characteristics and applications of
flash metakaolins. Appl. Clay Sci. 83-84, 253–262.

Schilling, M. (1990). Effects of sample size and packing in the thermogravimetric analysis
of calcium montmorillonite STx-1. Clays Clay Miner. 38 (5), 556–558.

Sesták, J., P. Holba, and Z. Zivkovic (2014). Doubts on Kissingers method of kinetic
evaluation based on several conceptual models showing the difference between the
maximum of reaction rate and the extreme of a DTA peak. J. Min. Metall. Sect. B
Metall. 50 (1), 77–81.

Sharp, J. H., G. W. Brindley, and B. N. N. Achar (1966). Numerical Data for Some
Commonly Used Solid State Reaction Equations. J. Am. Ceram. Soc. 49 (7), 379–382.
Bibliography 74

Shehata, M. and M. Thomas (2002). Use of ternary blends containing silica fume and
fly ash to suppress expansion due to alkali–silica reaction in concrete. Cem. Concr.
Res. 32, 341–349.

Shvarzman, A., K. Kovler, G. Grader, and G. Shter (2003). The effect of dehydroxyla-
tion/amorphization degree on pozzolanic activity of kaolinite. Cem. Concr. Res. 33 (3),
405–416.

Silcox, G. D., J. C. Kramlich, and D. W. Pershing (1989). A Mathematical model for


the flash calcination of dispersed CaCO3 and Ca(OH)2 particles. Ind. Eng. Chem.
Res. 28, 155–160.

Slade, R. and T. Davies (1991). Evolution of Structural Changes during Flash Calcina-
tion of Kaolinite. Mater. Chem. 1 (3), 361–364.

Slade, R., T. Davies, and H. Atakül (1991). Flash calcination of kaolinite: mechanistic
information from thermogravimetry. J. Mater. Chem. 1 (5), 751–756.

Slade, R., T. Davies, H. Atakül, and R. Hooper (1992). Flash calcines of kaolinite: effect
of process variables on physical characteristics. J. Mater. Sci. 27, 2490–2500.

Snellings, R., L. Machiels, G. Mertens, and J. Elsen (2010). Rietveld Refinement strategy
for quantitative Phase analysis of partially amorphous zeolitized tuffaceous rocks.
Geol. Belgica 13, 183–196.

Stoch, L. (1984). Significance of structural factors in dehydroxylation of kaolinite poly-


types. J. Therm. Anal. 29 (5), 919–931.

Stoch, L. and I. Waclawska (1981). Dehydroxylation of kaolinite group minerals I.


Kinetics of dehydroxylation of kaolinite and halloysite. J. Therm. Anal. 20, 291–304.

Szekely, J. and M. Propster (1975). A structural model for gas-solid reactions with a
moving boundary -VI. Chem. Eng. Sci. 30, 1049–1055.

Takkinen, S., J. Saastamoinen, and T. Hyppänen (2012). Heat and mass transfer in
calcination of limestone particles. AIChE J. 58 (8), 2563–2572.

Tavman, I. (1996). Effective thermal conductivity of granular porous materials. Int.


Commun. Heat Mass Transf. 23 (2), 169–176.

Thomas, M., M. Shehata, S. Shashiprakash, D. Hopkins, and K. Cail (1999). Use of


ternary cementitious systems containing silica fume and fly ash in concrete. Cem.
Concr. Res. 29 (8), 1207–1214.

Tikalsky, P. J., M. V. Huffman, and G. M. Barger (2001). Use of Raw or Processed


Natural Pozzolans in Concrete. Am. Concr. Inst., 1–24.
Bibliography 75

Tironi, A., M. A. Trezza, A. N. Scian, and E. F. Irassar (2012). Kaolinitic calcined clays:
Factors affecting its performance as pozzolans. Constr. Build. Mater. 28 (1), 276–281.

Versteeg, H. and W. Malalasekera (2007). An introduction to computational fluid dy-


namics (Second ed.). Essex: Pearson Education Limited.

Weber, J. N. and R. Roy (1965). Dehydroxylation of kaolinite, Dickite and Halloysite:


Heats of reaction and kinetics of dehydration at PH2O=15psi. Am. Mineral. 50,
1038–45.

Wen, C. and S. Wang (1970). Thermal and Diffusional Effects Solid Gas Reactions. Ind.
Eng. Chem. 62, 30–51.

Yin, C., K. Kaer, L. Rosendahl, and L. Hvid (2010). Co-firing straw with coal in a
swirl-stabilized dual-feed burner: modelling and experimental validation. Bioresour.
Technol. 101 (11), 4169–78.

Potrebbero piacerti anche